IT-Compendium

1. Computer Architectures 314 7.1.1 LTO Drives 7.1.2 dLT/SDLT Drives 7.1.3 ADR Drives 1.1 intel - Based Computer Systems 7.2 Helical Scan Recording Method 1.1.1 processors 7.2.1 exabyte Drives 1.1.2 Computer Buses 7.2.2 dAT Drives 1.1.3 Hard Disks 7.2.3 AIT Drives 1.1.4 monitors and Graphics 7.3 other Recording Methods 1.1.5 operating Systems 7.3.1 QIC Drives 1.1.6 Standard Applications 7.3.2 mainframe Drives 1.1.7 Special Applications 7.4 Software for Storage 1.1.8 network Applications 7.4.1 Sotware 1.2 RISC Computers from Sun and transtec 7.4.2 HSM Software 1.2.1 processors 1.2.2 Computer Buses 1.2.3 peripheral Buses 8. Media 341 1.2.4 operating Systems 8.1 CD-ROM Drives 2. operating Systems 317 8.2 CD-R and CD-RW Drives 8.3 dVD-Drives 8.4 magneto-Optical Drives 2.1 Windows 8.5 outlook 2.1.1 Windows 2000 2.1.2 Windows XP 2.1.3 Windows 2003 9. main Memory 343 2.1.4 Windows Small Business Server 2003 9.1 memory Technologies 2.2 unix-Operating Systems 9.1.1 non-Voltatile Memory 2.2.1 Linux 9.1.2 dynamic Memories 2.2.2 SunOS/Solaris 9.1.3 Static Memories 2.3 Computer Viruses 9.2 modular Structural Forms 3. Cluster 323 9.3 Terms relating to Memory

3.1 High Performance Computing Cluster (HPC) 10. Communication 346 3.2 High Availibility Cluster (HA) 10.1 definition 4. Storage Buses 325 10.2 prerequisites for Successful Communication 10.3 Communication Types 10.3.1 one-way Communication 4.1 introduction 10.3.2 Two-way Communication 4.1.1 parallel versus serial buses 10.3.3 interactive Communication 4.2 SCSI-Interface 10.3.4 open Communication 4.2.1 parallele SCSI 4.2.2 Serial Attached SCSI (SAS) 4.3 11. Standards 347 4.4 SSA 11.1 definition of Standard 4.5 ATA (IDE) Interface 11.2 Types of Standards 4.6 uSB Interface 11.3 important Orgaisations for Standardisation in IT 4.7 Fire Wire 1394 11.3.1 international Standardization Organization (ISO) 5. Hard Disks and RAID 331 11.3.2 international Electronical Commission (IEC) 11.3.3 international Telecommunication Union (ITU) 11.3.4 european Telecommunication Standards Institute (ETSI) 5.1 introduction 11.3.5 institute of Electrical and Electronical Engineers (IEEE) 5.2 Hard Disks Drives 11.3.6 ANSI 5.2.1 Capacity 5.2.2 Storgae Interfaces 5.2.3 Form Factor 12. The OSI Reference Model 348 5.2.4 performance Features 12.1 introduction 5.3 Semiconductor Disk Storage (RAM Storage) 12.2 Structure 5.4 Drives 12.3 Layers 5.5 Removable Disk Drives 5.6 RAID 5.6.1 RAID Level 13. Transmission Methods and 5.6.2 Criteria of RAID Selection Techniques 349 5.6.3 RAID Implementation 5.6.4 ide RAID 13.1 introduction 13.2 Signal 6. Storage Networks 336 13.3 Synchronism 13.4 Character Transmission 6.1 introduction: Storage Centralisation 13.5 operation 6.2 (SAN) 6.3 iSCSI 14. personal Area Networks - PANs 350 6.4 network Attached Storage (NAS) 14.1 definition 7. magnetic Tape Storage 338 14.2 Transmission Methods 14.2.1 irDA (Infrared Data Association) 7.1 Linear Tape Recording Method 14.2.2 Bluetooth

312 transtec = ttec in the Netherlands and Belgium p 17.8.4 p 17.8.3 17.8.2 17.8.1 digital Line (DSL) Subscribe 17.8 17.7.6 17.7.5 17.7.4 digital Data Transmission ina WAN 17.7.3 data Transmision ina 17.7.2 WAN 17.7.1 d igitalDSL) Line( Subscribe 17.7 p 17.6.4 17.6.3 17.6.2 17.6.1 17.6 17.5.2 17.5.1 data P 17.5 data Switching ina WAN 17.4.2 17.4.1 17.4 u 17.3.4 17.3.3 p 17.3.2 17.3.1 d 17.3 17.2 17.1 17. D distributed Queue d 16.2.1 m 16.2 16.1 m 16. m 15.10 15.9.4 15.9.3 15.9.2 15.9.1 15.9 15.8.4 15.8.3 15.8.2 15.8.1 15.8 p p 15.7.2 15.7.1 15.7 15.6.2 15.6.1 15.6 15.5.10 15.5.9 15.5.8 15.5.7 15.5.6 15.5.5 15.5.4 15.5.3 e 15.5.2 m 15.5.1 e 15.5 15.4 15.3 15.2.2 d 15.2.1 15.2 15.1 15. Wide Area - Networks WANs Local Area -LANs Networks Access Control inNetworks Attacks and Targets ofAttack in Security WANs Asynchrounous Transfer M X.25 andFrame Relay I Serial Analogue Data Transmission in WANs Circuit-Switched Networks Leased Lines WAN-Technologies Asynchronous Transmission M Synchronous Transmission M Transmission M Line Switching P Routing The Transmission Control P The I Adressing ina WAN Transmission Technologies for UpaM Setting Fibre- Copper-Based Cabling ConceptsBackbone Structured Cabling P Hardware- Switch Bridge Hub Repeaters Hardware-Active Components Standards FDDI Token Ring Ring Topologies Auto-Negotiation 1000 BASE-LX 1000 BASE-SX 100 BASE-T 100 BASE-FX 100 BASE-TX 10 BASE-FL 10 BASE-T 10 BASE-2 10 BASE-5 Token P CSM Accessing thernet Standardsthernet and802.3LAN thernet acket- andCell-Switched-Neworks rotection ofData Transmission rotective M rotocols ina WAN rotocols rotocols/Standards ser Datagram P efinition efinition efinition AC-Adresses anagement ANs etropolitan Area Networks A/CD nternet-P ptic-Based CablingOptic-Based acket Switching nterfaces ina nterfaces WAN assing rotocols assive Components easures andStrategies rotokoll (IP) ethods ina WAN rotocol (UDP) ual Bus(DQB) ode (ATM) rotocolP) (TC ethods ethods

AN transtec = ttec in the the in ttec = transtec 360 352 359 N etherlands and Belgium and etherlands data Communication 20. m 19.4 19.3 19.2 i 19.1 19. 18.7 Expansion network 18.6 18.5.3 18.5.2 18.5.1 Distributed with Core networks 18.5 18.4.1 18.4 18.3 18.2.5 18.2.4 18.2.3 newStrategies for Networks 18.2.2 i 18.2.1 18.2 18.1 18. 17.9.4 iSDNAdapters 17.9.3 17.9.2 17.9.1 17.9 dimensioning 24.5 24.4 online Technology 24.3 24.2.3 o 24.2.2 p 24.2.1 24.2 24.1 uninterruptible P 24. 23.4 digital Cameras 23.3 23.2 23.1 m 23. p p 22.3.2 22.3.1 p 22.3 22.2.2 22.2.1 22.2 m 22.1.4 22.1.3 unitsofM m 22.1.2 22.1.1 o 22.1 PCs network 22. Computers network (NCs) 21.6 pCsand Workstations 21.5 21.4 21.3 21.2 21.1 21. 20.1 LAN Core Solutions Terminals Bar-Code Readers Scanners Keyboards Résumé forA NewDirection Corporate Networks Standard Network Basic Network upaDistributed FabricSetting inaNetwork XRN-Components Current Solutions Availability Security andFlexibility Scalability Cost Control CoreStructure ofaNetwork Firewalls Routers Analogue M Hardware ina WAN Shutdown I Alarm, Line I Technologies Video Conference Systems Video P Voice Applications The Technologies The Technologies LCD Graphic Standards Types of M Windows-Based Terminals (WBTs) X-Windows Terminals Alpha Terminals M (Keyboard,KV Video, M ntroduction nput D roblemes, Causes andEffects rinter P rinters ros andCons ofthetwo technologies ffline Technology onitor Requirements onitors ice and Trackballs utput D ultimedia -P nteractive Technology rojectors rocessing nterfaces, Network-Widenterfaces, Communication rotocols easurement and Term onitors evices odems evices ower Supply ouse) Switching 377 374 379 388 386 380 382 313

Compendium 1. Computer Architectures

The following sections contain a brief overview of Intel-based computer systems was originally developed by IBM, Compaq and Hewlett-Packard. PCI-X allows a as well as the most important RISC workstations available today from Sun and slot to operate with 133 MHz and additional slots with 66 MHz or 100 MHz, thus transtec. achieving a throughput of up to 1 GB/s, which is double the throughput of the The CPU is the central and most important component of a computer system, and PCI. An advantage of the PCI-X is the reverse compatibility, which means the PCI- it is often the most expensive part. X cards can be used in PCI systems and vice versa. Another increase in the data Currently, the most important subject for processors is the transition from 32 to throughput is planned with the PCI-X 2.0. The maximum transfer rate is to be 64-bit architectures. Most RISC architectures have already taken this step, for increased to 2.1 or 4.2 GB/s using double data rate and quad data rate. example SPARC. The Intel IA64 architecture went to 64 bit with the Itanium in the summer of 2001. The follow-up model Itanium 2 has been available since PCI Express the end of 2002. AMD has been supplying the Athlon 64 (desktop version) In the future, PCI-X is to be replaced by PCI Express - the company that started and the Opteron (server version) since the 2nd quarter of 2003. A point worth under the name 3GIO (3rd Generation I/O). As a serial „rail bus“, the PCI Express mentioning in connection with the AMD 64-bit architecture is the full 32-bit can be simultaneously scaled with up to 32 rails that can travel in both directions. compatibility. As a result, 32-bit operating systems and applications also run with With the planned transfer rate of 2.5 GHz, each line is to be equipped with 250 very good performance. An important factor though is not only the availability of MB/s in both directions, which means a maximum of 2 x 8 GB/s at full capacity. the hardware, but also that of the right operating systems. Microsoft‘s Windows PCI Express should be available as of 2004. Server 2003, the successor of Windows 2000, is available in the Enterprise version with support for the IA64 architecture. A version for AMD‘s 64-bit architecture InfiniBand is in preparation. In the Linux sector kernels with respective 64-bit support are Another high-performance interface for servers is the InfiniBand from the InfiniBand available for both architectures. However, only when there are applications Trade Association. This is the result of the merger of the Future and NGIO groups. available with 64-bit support, can the user take full advantage of the performance Unlike PCI-X, InfiniBand does not offer downward compatibility, but it thereby of the 64-bit architectures. presents the great advantage of being able to offer new functions. The basic concepts of the InfiniBand are extremely similar to those of Fibre Channel: Systems 1.1 Intel-Based Computer Systems are connected to an I/O „fabric“ using one or more Host Channel Adapters (HCA). With Target Channel Adapters (TCA), storage or network controllers are 1.1.1 Processors connected to this fabric. The InfiniBand is addressed with IPv6 addresses, which means a virtually infinite number of units can work together. The can Processors from Intel are mainly used in PCs. be scaled up to 10 GB/s. Using copper cables, the cabling can reach a maximum Whilst the choice of was relatively straightforward during the era length of 17 m, and using glass fibre, up to 100 m. This design allows considerably of the Pentium and Pentium Pro, it is becoming increasingly more difficult to smaller and systems to be built with InfiniBand, as bus systems are select the correct motherboard due to the different processor‘s designs and the not needed in the server. You can find more information on the InfiniBand TA wide variety of chip sets that are available. There are now numerous possibilities homepage at: www.infinibandta.org. for processor slots and the life span for technologies is decreasing tremendously. For this reason, it is practically impossible to equip systems older than one year AGP with, for example, the latest processors. Therefore, when planning, it should be The Accelerated Graphics Port is an individual graphics card slot developed by assumed that only main memory and hard disks can be easily updated at a later Intel. Conventional graphics cards use the graphics memory to partially store time. In most cases, the performance of today‘s processors does not need to be textures. The graphics memory on the cards was developed to display increasing improved by an update. amounts of data at increasingly high resolutions. However, this memory is not used in standard applications. The AGP concept allows the graphics card to 1.1.2 Computer Buses directly access texture data stored in the main memory. A graphics application can optimally select texture data according to size, colour depth, and resolution, The choice of a suitable bus system is fairly simple because there are only two and can transmit this information to the AGP card in the main memory. The major bus types available for PC systems: PCI bus and AGP bus. Other bus systems main-board chip set and AGP interface exchange the data at 66 MHz with 32 such as ISA, EISA or Microchannel are no longer being used. Nearly every system bits, which corresponds to a transfer rate of 266 MB/s. Additionally, the AGP card available today has PCI slots and an AGP slot. The AGP bus supports only graphics supports the so-called x2 mode or x4 mode. In the x2 mode, data is transmitted cards. on both edges of the clock signal so the data transfer rate increases to 533 MB/s, The following section contains a brief survey of the bus systems currently used in four times the data transfer of the PCI bus. A further increase is possible with 4x. PCs and servers. The AGP 4x also includes the additional strobe signals AD_STB0# and AD_STB1#, which form differential line pairs with the standard strobes. These four strobes PCI Bus are then able to work at 133.34 MHz. Even with the AGP 4x, the transfer can be The first PCI bus version already operated at 32 bits with a clock rate of 33 MHz. successfully carried out with the falling and raising edges of the AD strobes. The The PCI bus can support a maximum transfer rate of up to 133 MB/s. The 64-bit strobe signals’ 133.34 MHz produce four cycles per AGP clock, which effectively version supports a maximum transfer rate of 267 MB/s and a crucial advantage means 266.67 MHz and a theoretical band width of 1017 MB/s. of the PCI bus is that it is not dependent on the processor type. Thus, the PCI AGP is downwards compatible with PCI: PCI bus masters, for example bus can also be used with „non-Intel processors“, e.g. in computers with SPARC Framegrabber cards, can write data onto the AGP graphics card with the PCI architectures. The PCI bus is supported by the PCI Special Interest Group whose bus. The high bandwidth at which the AGP graphics card can access the system’s members include all major manufacturers of processors, circuit boards, hard disks, RAM required changing to SDRAM with 133 MHz or DDR modules for RAM controllers and graphics cards. technology.

PCI-X I2O Soon the performance capability of the PCI bus will not be sufficient to meet I2O (Intelligent I/O) is another product which was developed by Intel. I2O is based the increasing demands being placed on the performance of servers. For this on the Intel i960RP I/O processor, which contains a complete subsystem on a reason, the PCI Steering Committee has created the PCI-X specification. PCI-X single chip. Due to the on-chip PCI-to-PCI bridge, the i960RP can be connected

314 transtec = ttec in the Netherlands and Belgium a specific usually depends on the application programmes to be programmesto application the on depends usually systemoperating specific a of here.given be selection Thenot will systemsoperating for Recommendations 1.1.5 Operating systems prices. higher the of because just monitors quality better the fromdeterred be not do component, other any than longer standardscomputer future meet to continues monitor the Since damaged. be monitor,can the byit supported than frequency card.graphics the of specifications performance the supports also fidelity,and etc.)colour high radiation, electromagnetic low resolution, (high workplace ergonomic an requirementsof the all meets which monitor suitable a select to important is Furthermore,card.it the on installed memory video the and processorgraphics the by determined mainly is card graphics the of software. quality driver properThe the using when achieved be only can monitor the of frequency maximum the and cardgraphics the of with resolutioninstanceWindows forXP). maximum The(as system operating the with provided is or systemoperating the foravailable commercially is software driver that ensure to important is it cardgraphics a choosing When 1.1.4 M capacity. storage of megabyte per price falling rapidly and advancementtechnological the to due is which GB, 40 at meanwhile is size used.system entry operating The EIDE an be should it whether decide must user the disk, hard a purchasingBefore 1.1.3 Hard disks expansion. foralternativeversatile and simple a are slots card cards,the expansion for slots internal any with equipped not normally are and interfaces external of number limited a only havegenerally computers of types these computers.Since laptop and notebook with used primarily are cards PC However,slot. CardBus new older the the with (e.g.Fast performance higher hardwarewith sophisticated more use to notebooks allows bus the of that to similar 132 to up of ratestransfer with to similar bus 32-bit performance. Thehigher and compatibility better consumption,power reduced as improvementssuch certain has which other.the each of of development latest The type A disks.hard in example for type type thickness: card in rows.only two externally differin Theyarranged holes pin 68 has strip socket their and mm, 54 x mm 85.6 of dimensions surface standardisedhave cards PC adapters. LAN and cards,modems interface the accordancewith in used and designed be could hardwareother so expanded graduallycards. was standard Thememory strictly ( the by established was The PC Card I today.used not practically is D processor.this for specially developed software requires this although system, entire the of performance improvesthe thus and interruptsprocessing from CPU the to directly An important consideration for operating systems is the size of the main the of size the is systemsoperating for consideration important An themselves.users the by made best is decision this all, After used. PCMIA espite these advantages, these espite C card interface is a standard for credit-card-sized expansion cards,which expansion credit-card-sizedfor standard a is interface card PC . The size of a hard disk depends on the application and the and application the SATAon , depends disk hard a of SCSI. size Theor with a 5 mm thickness for cards, and type type cards,and for thickness mm 5 a with II onitors and Graphics ) in September 1990. The first products that used this standard werestandard this used that products 1990. first TheSeptember in ) A standard, which means that older that means which standard, PCMIA nterface

then frees up the up frees then chips. i960RP additional anyThe without bus PCI

with a 3.3 mm thickness, mostly used for memory cards; memory for used mostly thickness, mm 3.3 a with I with bus master capability and a frequency of 33 of frequency a and capability master bus with PCI bus, but for credit-card sized peripherals. sized modern credit-cardThis for but bus, PCI E P thernet). thernet). The ersonal

was not able to get a foothold on the market and market the on foothold a get to able not was I2O

B/s. Thus, the performance of the CardBus is CardBus the of MB/s.performance the Thus, C card usually fits into two type type two into fits usually card III omputer I f a graphics card is run at a higher video signal video higher a at run is cardgraphics a f C card standard is downwardcompatibleis standard card PC A slots do not support new support not do slots PCMIA M C CardBus specification describes a describes specificationCardBus PC

C card standard is the CardBus the is standard card PC emory emory with a 10.5 mm thickness, thickness, mm 10.5 a with III A cards can be used in used be can cards PCMIA C C card standard, such as such standard,card PC ard I nterface nterface transtec = ttec in the the in ttec = transtec slots on top on slots II

A ssociation I/O C cards.PC Hz MHz PC

N etherlands and Belgium and etherlands projector would be advisable. be wouldLC projectoran D of purchase I applications. multimedia complement to useful are cards sound and applications, diskettes. on run to large too is data generated of amountusually, because,the used are packages presentationgraphic or database if considered be also should burner CD or user.tape office a typical of purchasethe Thefor sufficient rule a as are graphicsonboard programmes.presentation graphicswith systems standard The word-processing,and include spreadsheetdatabase, applications Standard 1.1.6 Standard Applications interface. graphical a using when the on depend WindowsXP 128 Windows2000, RA between forth O as RA of amount minimum the then today.applications for sufficient RA recommended systems,the Forsome dimensioned. generously manufacturer.system operating the of recommendationsexceeds the considerably needed actually properly.them run to needed memory I limits. performance its reaches server network powerful most the workstations,even connected of growingnumber continuously a With system. ID RA a using when evenguaranteed be cannot security data otherwise, (UP S), supply powercontinuous,uninterruptible a with system. ID RA a of use the is solution optimal An loss. data against overallprotectionprovide not does measure However,workstation. a into this or server the intoincorporated be either can which medium backup suitable a with equipped be should servers Network dependability.and extendability for designed be tohave disks hard and enclosureunits, processors,ventilators, supply power an in element central a requirementsfor the meet should server a of overalldesign The availability.and reliability affect severely also may more,it functions. is What server the of performance the limits severely it recommended,as not is operation However,of type this workstation. a as used be also can server the that operation. means Thisserver non-dedicated permit solutions server Some card. network high-performance a medium. storage the of made is use optimal that so speed, same the at run controller disk hard the and processor, the the that important more is it server a sufficient.using is When monitor VGA role.lesser a Therefore,playssimple output a graphicsserver, the network a On Applications1.1.8 Network medium. backup a requirement- card.graphics and processor,the of RA area the in particularly performance, higher of be to has system the Accordingly,applications.of standard equipment than systemsthe (D publishing desktop or processing (CAD designcomputer-aided multimedia, as such applications Special 1.1.7 SpecialApplications f presentations are to be given using a using given be to arepresentations f n high-load networks, it should therefore be determined, whether or not the load the not or whether determined,therefore be should it networks, high-load n therwise the processor will spend too much time swapping data back and back data swapping time much too spend will processor the therwise M icrosoft icrosoft T infrastructure. All the component such as chip set, main memory, main set, chip as such component the All infrastructure. IT , it should be 256 be should it , O DV burners have become standard. They can then also be used as used be also standard.then becomehave can Theyburners DV ffice are to be used, then a minimum of 64 of minimum a then used, be to are ffice nix variant used. However,used. 64 variant least Unix at and the hard disk. For applications such as Windowssuch Forapplications disk. hard the and M -R I n these sectors sectors these n drives perform very well when working with graphical with working when well very perform drives OM B is needed to reach an acceptable processing speed.processing acceptable With an reach to needed is MB D ue to the low prices, the main memory should always be always should memory main the prices, low the to ue

I n any case, the server system should be provided be should system server the case,any n B. ForMB.

I f or WindowsWindows98 should be 32 be should M drives have meanwhile become a basic a become meanwhilehave drives DV ) require more sophisticated computersophisticated morerequire TP) nix operating systems the requirementsthe systemsoperating Unix I C during lectures or training sessions, the sessions, training or lectures during PC t is possible that the amount of memory memory of amount the that possible is t is 16 is M

B. However,MB.not is this MB. I t is also very important to use to important very also is t I f software packages such packages software f B should be available be should MB

B should be available.be should MB M e is to be used, be to is e ), graphics ), T or NT M 315

Compendium on a server can be reduced by the use of several servers, or by a division of the systems, such as “Blade 1000” and “SunFire 280R”. The PCI models support the network into smaller sub-networks linked with bridges and routers. PCI bus specification in a 64-bit, 66 MHz configuration. They can theoretically reach a throughput rate of 528 MB/s. The availability of expansion cards is The configuration of a workstation computer depends on the tasks it has to currently still very limited because most card manufacturers only include drivers for perform. A workstation computer does not necessarily need its own hard disk, the Windows operating systems with the cards. Solaris requires special drivers. The since it is possible to boot the computer directly from the server‘s hard disk if the PCI bus has completely replaced the SBus in all computer systems. network card has a bootROM included. The UPA bus serves as a processor bus and is used in all UltraSPARC systems. Up to 64 UPA slots are available, depending on computer model. 1.2 RISC Computers from Sun and transtec The Ultra Port Architecture (UPA) uses an internal Crossbar Switch, which links and co-ordinates the CPU, the memory, and the I/O components of the system. 1.2.1 Processors With its throughput of up to 2.4 GB per second, the UPA shows an excellent performance. The UPA bus uses the 64-bit UltraSPARC processor with integrated Sun Microsystems is introducing its third generation of UltraSPARC 64-bit processing very effectively. microprocessors, the UltraSPARC III processor, which was developed for the fast-growing and its corresponding data load. It sets new standards 1.2.3 Peripheral Buses for scalability, performance and data throughput and is the heart of the new generation of products from Sun and transtec. Now, Sun Microsystems can offer The new models from Sun are equipped with FireWire (IEEE 1394) and USB. The three types of processors: Ultra IIe, UltraSPARC III and the new UltraSPARC IIIi Universal Serial Bus is primarily used to connect the keyboard and mouse. In the which are available in different clock speeds. future, a fibre-channel will be used for internal hard disks and Sun’s UltraSPARC processors are available in 360, 400, 450, 480 MHz and have external storage arrays. An Ultra SCSI interface is available for external peripherals. an L2 cache with up to 8 MB. These processors are used in all UltraSPARC models Network connection is established using a 100 Base-T interface. For all high-end and in an Enterprise server up to 106 CPUs can be installed (Sun Fire 15K). The graphics applications, the CREATOR Frame Buffer, which is plugged into a UPA models with Ultra IIe are only available in single-processor versions with 500 MHz. slot, provides high performance. The CREATOR card combines both the large The new UltraSPARC III processor has clock speeds of 600, 750, 900 and 1050 linear memory of a frame buffer and the direct access to the 128-bit UPA system MHz. Sun’s roadmap envisages clock speeds of up to 2.1 GHz. transtec offers a architecture with the powerful Visual Instruction Set (VIS) of the UltraSPARC CPU. wide range of alternatives, which are either identical to the original Sun model or differ from it only slightly. 1.2.4 Operating systems

1.2.2 Computer Buses Solaris, which is known for its maturity and stability, is the main operating system used on SPARC computers. The current version, Solaris 8, is fully 64-bit compliant Sun systems use two different bus systems for expansion cards: and meets the most stringent security demands as well as providing optimum There are two versions of the SBus available: a 32-bit version with a transfer rate performance for Java applications. Further information on Solaris can be found in of 40 MB/s and a 64-bit version with a transfer rate of 100 MB/s. The system the chapter Operating Systems. automatically recognises the type of card installed and selects the appropriate The Linux operating system is also available, but is not yet widely used. speed accordingly. The second bus system is the PCI bus, which can be found in all of the current models, for example Ultra60, Ultra450, and in the new Ultra III

316 transtec = ttec in the Netherlands and Belgium 2. them. change not can user the that so set be also can configurations workplaces. respective The the at installed automatically be will software user's the and all workplaces Windowsat configurations2000 same the exactly have will user the that filed.centrally means be This can groups or users Active the of configured.help the With I expandable and configurable the and management configurationcentralised features include under is there as domain per users approx.20-40,000 of limit a longer no is there structured, more just not is administration user the Windows2000, Active the in containers in hierarchically managed and configured resources. system and network of management and to access simplifies that platformcentral Administration: line. command the in implemented been has mechanism improvedrepairadditional, An mode. boot Repair stage.later a at installed be can Active the for service Configurethe controller,of Yourthe wizard,domain help not. the Withor Server the as used be should the server Windowsif 2000 determine to necessary not is Windows2000. to comparison Windowsin seldom very rebooted be must it that to is improvementmade Windows2000 of CD bootable a using systemoperating an without I server.8-processor an on up set be can processors 4 with each Servers virtual 2 example for servers, multi-processor on run be can systemoperating the of servers. instancesseveral Thus, virtual of support the featureis important Another management. load work forcontrolprocess and servers 4 for clustering failover over AdvancedServer: Windowsfunctions 2000 memory. main GB 64 and processors 32 to up The available.also are servers two for clustering failover and the functions, the Server Windows2000 Server The Windows version. server this in contained already the replaceTerminalWindows which services, processors.four and The Windowsmemory main GB 4 max. for hardwaresupport Server Windows2000 processors.two and memory main GB 4 to up supports Windows2000 versions:four in Windows2000 hardwaremanagement. and storage as wellexpandability, as with Windows compared functions, featuresand new several of addition the by expanded M 2.1.1 Windows 2000 2.1 Windows and Unixoperating systems. The following provide ofthemostcommon M sections overview abrief memory. the With and Scalability ntelliM of Windowsnstallation 2000: icrosoft Windows 2000, referred up until now as nowWindows until icrosoft referred up Windows2000, M Windows2000 O icrosoft Windows 2000, which simplifies the installation process.Another installation the simplifies icrosoft which Windows2000, . M p to 8 GB main memory and 8 processors are supported. supported. areprocessors 8 and memory main GB 8 to Up irror technology allows Windows 2000 workplaces to be centrally be to allows technologyworkplaces Windowsirror 2000 perating Systems echanisms: nlike in the in Unlike M NT icrosoft icrosoft T. These concern the areas of administration, scalability and scalability NT.administration, of areas the concern These Extendability: P

nlike the WindowsUnlike M rofessional Enterprise icrosoft Windows 2000 implements Active icrosoft implements Windows2000 M

atacenter Server Datacenter Windows 2000 is equipped with an improved,protectedan with equipped is Windows2000 icrosoft Cluster Server, two or more servers can operate in operate can servers more or Server,two Cluster icrosoft irectory (directory service especially for user administration) user for especially service (directory Directory is the successor of the Windowsof successor the is M U C). (MMC). Console anagement ser NT. The corresponds to the toWindows corresponds E Windows 2000 supports up to 64 GB of physical of GB 64 to up supports Windows2000 M dition will continue as continue will dition Windows 2000 is installed in the computer the in installed is Windows2000 anager for anager Windows

SB support has also been implemented in implemented been also has support USB T server, when installing it server,Windowswhen 2000, NT represents the top end, supporting supporting end, top representsthe irectory, the configuration presets forpresets configuration the Directory,

load balancing (with up to 32 servers) servers) 32 to up (with balancing load IP I T 4.0 are TerminalNT edition, Server t offers the following additional following the offers t .

P T, users can be organised, be NT,can users Windows 2000 AdvancedWindows2000 lug & lug

T Server and offers and Server NT

NT Workstationand P lay is a new featurenew a is lay T 5.0, has been has 5.0, NT irectory as a as Directory transtec = ttec in the the in ttec = transtec M icrosoft offers icrosoft I Directory. With n addition to addition n

NT.Further icrosoft icrosoft N etherlands and Belgium and etherlands and digital e-mail signatures. e-mail digital and log-ons user in security the increases Cards,which Smart supports Windows2000 Furthermore,5. Kerberos accordancewith in introduced is authentication (SSL). Layer Sockets Secure in use Security: Network drivers. certified of installation files.system operating critical preventsdeleting Functions: Security and WindowsWindows98 2000. Driver difficulty.any without computers Hardware easily. manner.organised 2000/NT, the With level. directory or level file on data sensitive of encryption users. the Theto available space disk of amount maximum the define to quotas implements also now NTFS productivity.increasingthereby workload, the share can servers the fails. servers the of one if interruption without operation maintain to as so other each way.monitorinterlocked can an devices the Thus, memory. of management computer’sefficient and potential the to interface user the adapting activities, systemfor downtime utilizing The faster. programsstart means also earlier Windows versions.with function than same fasterThe 34% to called access.rapid for Togetherdisk so- hard the the with on files start the arranges XP hardware,and software Windowssame the with starts several up.After started is computer the moment the clear improvementunder becomes Windows XP enhanced.significantly first been The also has used. universallyperformance The be can providesfunctions Windows XP playbackrecording, and processingthe or scanners cameras,digital as the improvementgreatin a been also has There background. the in run to continue applications open the user,all in because change a is there the from file a downloading person’sother thatrunning. example,for keep means, Thisapplications the while simply and quickly changed user.be each forcreatedcan users The one use to persons severalfor up.called when again use forready immediately are applications the all because nothing,for run tohave longer no laptop the of batteries enhancing I type. application organized.and clean taskbar the keeps programand Webavailable.alwaysmail browserare The e- default the and first,displayed access.programsare most-usedeasy five The The developed. been have capabilities additional of multitude a design, new practice. relativelylittle with users foreven operationintuitive permitting design,improved, significantlyclear its with impresses WindowsXP 2.1.2 Windows XP has a series of further novel featuresnovel further of series a design, has the WindowsXP to addition n enhanced multitasking performance multitasking enhanced menu Start redesigned P M

M refetching icrosoft icrosoft Windows ) is intended to enable identical driver software to be used in used be to software driver identical enable to intended is DM) (W odel FS , the distributed structures of files and data on Windowsdata and files of structures distributed the system, file DFS NetWareand M user friendliness user anagement:

feature and an optimised network login, the system starts up starts system the login, network optimised an featureand The Security Service Service Security The

I I n this way, users can find files in the network much more much network the in files find way,can this users n n order to increase the operational security,operational the Windowsincrease2000 to order n

nix servers can be summarised and presented in an in presented and summarised be can servers Unix P lug & lug T 4.0, for example in the in example for 4.0, NT TFS expansion NTFS groups the most frequently used applications for applications used frequently most the groups I . Windows wakes from hibernation far faster,far the i.e.hibernation . from Windows wakes n Windows 2000, the SSL is expanded and Kerberos and expanded is SSL n the Windows2000, I nternet does not have to be discontinued when discontinued be tohave not does nternet C together or side by side. A separate account is accountseparate A side. by side or together PC

V cameras are automatically recognised and all recognisedand automatically are cameras DV P

lay allowslay I n addition, the expansion of the Windowsof expansion the addition, n P pen files are now grouped accordingtogrouped now are files Open rovider is supported by measures including measures by supported is ncryption File System) allows the FileallowsSystem) (Encryption EFS C cards to be used with portable portable with used be to cards PC I I ) is already in already is (SSPI) nterface hardwaresupport n addition, it only allows the allows only it addition, n T LAN NT D new taskbar grouping taskbar new

uring normal operation, normal uring M

anager and the and anager I n addition to the to addition n I t is also possible also is t . D

evices such evices

317

Compendium The Standby and Hibernate mode are of assistance for laptops. In Standby, the particularly sensitive information to be stored in encrypted form on the hard disk system powers down the monitor, hard disk and other devices, but continues to of the PC or on network servers. This dependably protects data including offline supply the main memory where the open documents and programs are stored. files and folders from unauthorised access. The new multi-user support of EFS After the system has been re-activated, work can be resumed where it left off, in allows multiple (authorised) users to encrypt and jointly inspect sensitive files. less than 2 seconds with relatively new laptop models. If the laptop enters the The Internet firewall (Internet Connection Firewall - ICF) integrated in Windows Hibernate mode, the data from the main memory are saved to the hard disk in XP automatically protects data from unauthorised access from the Internet. The compressed form and the computer is shut down completely. Startup from this function just has to be activated for the data communication or LAN connection mode takes a little longer, but the desktop state is as it was before Hibernate was wanted. Windows XP Professional uses 128-bit encryption for all encryption entered. A further improvement has been made to the power management operations. In addition, there are many further security features in Windows XP, functions introduced in Windows 2000. Hibernate and Standby now operate such as Kerberos V5 authentication support, Internet Protocol Security (IPSec), considerably faster and more reliably, saving Notebook batteries and increasing centrally defined security guidelines and a lot more. mobile operating time. With Remote desktop Windows XP allows users remote access to all their A series of features in Window s XP simplifies installation and administration of applications, documents and network connections on their desktop computers. the operating system and helps cut the costs for these tasks within the company. Through a secure, encrypted Internet or dial-up connection, users can link to the Just a selection of these features is given here. company PC and create a virtual session there with the customary Windows For companies wanting to install XP Professional on several PCs, Microsoft has desktop on the remote computer. Since all the programs run on the company PC provided sophisticated, tried and tested mechanisms for automatic installation. and only keyboard and mouse inputs and display outputs are transferred, the These allow unattended Windows XP installations to be pre- configured, enabling remote desktop is also ideally suited for modem or ISDN connections. the setup routine proper to run reliably and without user interaction. Operating Wireless networks (such as 802.11b-WLANs) are supported by Windows XP system images can be created with ease, even for PCs with different hardware, Professional as standard. Windows XP Professional recognises and configures and can be installed automatically across the network with the aid of the everything automatically as soon as a WLAN connection is established or cell-to- Windows 2000-compliant Remote Installation Service (RIS). cell roaming takes place. Certificates that are distributed or stored on smart cards Of interest to international companies and educational establishments: Windows as well as 802.1X authentication ensure the utmost transfer security (both in cable XP Professional supports the Multilingual User Interface (MUI): A PC can have and in wireless networks). a number of interfaces at a time (English, German, Spanish etc.), providing every Windows XP Professional recognises networks and their TCP/IP settings, so that user with the Windows desktop in the language he or she prefers. Administrators the network connectivity of the PC is established automatically and is can deploy software restriction policies to prevent unwanted applications from maintained during roaming. If the network does not provide for automatic running on Windows XP Professional PCs, thus restricting workstations to the configuration, Windows XP Professional allows working with alternative TCP/IP programs important for the company. In addition to the Group Policy settings settings. In this way a notebook user can draw the TCP/IP configuration provided for Windows 2000, there are numerous new ones provided for Windows dynamically from a DHCP server in the company network and work with static XP Professional for even more comprehensive, policy-based management through TCP/IP addresses at home. Active Directory. With the new Resultant Set of Policy (RSoP) Windows XP Professional, administrators have a powerful tool to plan, monitor and Windows XP is an uncompromising further development of the proven Windows troubleshoot the impact of various group policies on a specific user or computer. 2000 technology and therefore operates more stably than its predecessors. The The location-related group policies prove to be exceedingly useful for users often new Windows engine is based on the dependable 32-bit architecture of Windows en route with their notebook. If the user is at the office, the company group 2000, featuring a fully protected memory model. During installation, the setup policies apply. When the user is away or at home, however, he can employ the program of the operating system can incorporate driver, compatibility and security Windows XP Professional functions useful for single PCs or in small LANs, without updates from the Internet if the user chooses to install them, even if such have this requiring reconfiguration by an administrator or the user himself. become available after the Windows XP CD-ROM was issued. Windows Update The USMT (User State Migration Tool) allows easy migration of a user’s data keeps the system abreast of the latest changes. Service packs, troubleshooting and personal settings from the original system to Windows XP. Preferred desktop and drivers for new hardware devices can be downloaded from Microsoft on the settings, folder options of the Windows Explorer, Internet configurations and Internet easily and automatically if the user so wants. favourites, e-mail access, messages and addresses, network drives and printers, A large number of programs run under Windows XP that failed to run under specific folders etc. can be migrated with ease. Windows 2000. If a program can be run only under Windows 95 or Windows 98, most likely it can now also be run under the new operating system. Should an Windows XP offers universal support and end-to-end-integration for all digital older application nevertheless not be supported by Windows XP, such can be media. Transferring and recording, display and playback, archiving and sending - performed in a special compatibility mode presenting the characteristics of Windows XP offers context-sensitive support through every step and automatically earlier Windows versions. proposes all the tasks coming into question. Windows XP opens up audio, photo and video capabilities as never experienced before. The Media Player can create Windows XP gives top priority to the quality of device drivers since it MP3 files and play DVD videos via the PC through separate plug-ins. Windows XP contributes significantly to system stability. When a new component is installed, automatically recognises if a digital camera or scanner is connected to the PC. an indication is given whether Microsoft has tested and certified the driver. If a Windows XP identifies which files are in a folder and not only displays a non-certified driver has been installed and fails to operate correctly, the system thumbnail, but also proposes the most likely options for the respective media can revert to the earlier operational driver within a matter of seconds. By means type in the new taskbar: audio data, for example in the My Music folder, is of the side-by-side DLL support, multiple versions of individual components can played at a click of the mouse. run side by side, so that every application uses the version best suited to it. If The Remote Assistance integrated in the Help and Support Center of Windows failure is experienced despite the significantly enhanced driver handling, the XP enables colleagues or an in-house support centre to be called for help at any System Restore feature can restore the PC to a previous state and thus cancel all time. Instead of the problem being described to a specialist in words over the configuration changes made in the meantime, even after weeks have elapsed. phone, the expert can show the solution in a clear and intelligible manner on the user’s own screen by remote control. If the expert’s aid is enlisted via the online Great importance is attached to security in Windows XP. Sensitive files can be contacts of the Windows Messenger or via e-mail, he receives the authorisation to stored encrypted, passwords and user names in special areas are protected from view the user’s screen on his own PC to help find or even implement the solution. unauthorised access. The Encrypting File System (EFS) permits files containing The user can follow ever y step and retains full control at all times. The expert’s

318 transtec = ttec in the Netherlands and Belgium Application Verification: running. and up server the HardwareSupport: Server secure. I updates.systemoperating critical these install to when select patches.Administratorscan other and fixes security as updates,such systemoperating critical downloadsystematically Update controllers.domain office branch in caching credential and replication, synchronisation, efficientunreliable overmore even to WA thanks connections, N Active network. end-to-end secure a make to helping network, the of edge the tocore the from resourcesspan Services: Directory administrator’srequirements. the on based disabled or enabled is functionality server Lockdownwizard, Security run. be can services required only that ensure to setup during them. accessThe to used devices the or platform, or systemoperating their built,were applications the way the of independent applications other from invokefunctions that standards industry on XML WebServices: 2003: featuresof Windowsprincipal Server the of overview An partners. special through only available be Active the offer independently not but domain, Active an of members be can systemoperating this using server.Servers 2003, Web WindowsServer TwoRA GB Two-way(SMP) multiprocessing symmetric 2003, Web WindowsServer services. and hardware, software of package a offering the Windowsunder available be only will version This SANs foraccess direct Windowssockets: RA GB 64 (SMP) multiprocessing symmetric 32-way 2003, WindowsServer Non-Uniform Add Hot RA GB 32 nodes eight to up with Clustering (SMP) multiprocessing Eight-way symmetric 2003, WindowsServer RA FourGB Two-way(SMP) multiprocessing symmetric Standard 2003, WindowsServer available: be will scratch.versionsfrom following2003 or The WindowsServer further developed were functions the Server,featuresand manyComparedtoWindows 2000 new The 2.1.3 Windows 2003 Server problems. software hardwareand with support for costs the cuts hence and reduced be to centrecompany’s the support toown addressed support for calls of number the of allows toolWindows XP support ends.central comprehensivesession and This user’sthe for authorisation Assistance Remote issues. compatibility and corruption heap view,toinstance, fora with advance in tested nternet Firewall:nternet

M M M in 32-bit versions and 128 GB RA GB 128 and versions 32-bit in M RA GB 64 and versions 32-bit in M anagement: emory emory M icrosoft has been available since April 2003. April since available been has 2003 icrosoft WindowsServer M

M emory Access(NUM emory server’sThe

Active L Web services provide reusable components built components reusableprovide XML TheWeb services The automatic update provides the ability to ability the providesupdate automatic The Datacenter Enterprise Applications running on Windows Server 2003 can be can 2003 on Windowsrunning Server Applications

irectory security settings for users and network network and users for settings security Directory river verifiers check new device drivers to help keep help to drivers device new check verifiers Driver

E E dition is conceived especially for use as a web a as use for conceivedespecially is dition dition I

nternet firewall makes firewall nternet E E dition

E dition irectory is now faster and more robust,more and faster now is Directory

dition A) S 6.0 security settings are locked down locked are settings security 6.0 IIS

in 64-bit versions 64-bit in M

in 64-bit versions 64-bit in M

C expires automatically after the after automaticallyexpires PC irectory service. This version will versionservice. This Directory I nternet connection more connection nternet Datacenter

transtec = ttec in the the in ttec = transtec U

sing the sing

P rogram,

IIS irectory Directory N etherlands and Belgium and etherlands Windows and non- devices.WindowsWindows-basedand client locally.both for software capability the Terminalthis providesrun cannot Server P WindowsXP virtual a access can server.user Forthe a example, on running TerminalServer: intervention. administratorsystem requiring without shares network on time. this at existed they as folders the of status the determine later and contentsfolder network save can Copy: Shadow network. the on computer other any virtually fromcomputer a manage can administrators Administrationmode), Remoteas Terminalin known (formerly Services Administration: Server Remote shares.file creating and users new adding as tasks common such perform to easy it server, making the of management ongoing for interface M time. first the correctly configured and installed are components that server,access remoteensuring server, server, or print file a as such roles server various up setting of process the throughadministrators Configure Your Wizard:Server file. log a to shutdowns tracker.shutdown server new the using Server The additional functions of the of functions additional The products. balancer load external third-party by supported is and service balancing load 2003 the WindowsServer Session session. disconnected a to reconnect to users TerminalSession Services of set a using applications to available made features.is informationmanagementAdditionally, topology the memory and scheduling,thread settings, affinity processes,default application awarenessto 2003, system. WindowsServer Affinity ResourceTablethe Static the describes thatcalled Non-Uniform hardwaresupport. respective the havethat servers on only downtime.operate will featurecurrentlyThis involvesno and re-booting require not pool. does This memory normal the of part as applications and systemoperating the to available made and computer Add Hot organisations.acrossinformation identity synchronise with processes business integrationof the enables information, Active with files and databases directories, multiple frominformation integrate identity to companies helps M sites. multiple spanning clusters dispersed Geographically - - storagededicated with configurationscluster Single - differentconfigurations:of variety a in deployed be can clusters Server options. improvedprovidingscaling as well environment,as cluster dispersed geographically a hardwarein removing and adding for flexibility increased for and Service: Cluster the of functions additional The rofessional desktop and x86-based applications for applications hardwareWindowsthat fromx86-based and rofessionaldesktop M anage Your Wizard:Server etadirectory Services Support: Services etadirectory ultiple clusters on a storage area network (SAN) network areastorage a on clusters ultiple E Datacenter vent Tracking: M

emory: M This feature provides time-based network sharing. Administratorssharing. network featuretime-based providesThis emory Access(NUM emory The cluster service for Windows Server 2003, for WindowsServer service cluster The When using programsaccessWhenTerminal can user Server,a Hot Add Hot E dition supports up to eight-node clusters.eight-node providesto This up supports dition Administrators can report an accurate record of uptime accuraterecordof an report Administratorscan

nd users can recover accidentally deleted files or folders or files recoverdeleted accidentallycan users End Directory: M The emory allows memory to be added to a running a to added be to memory allows emory ConfigureThe Yoursteps wizard Server

Enterprise Remote With Directory. Datacenter Enterprise

M M

anage an Yourprovideswizard Server icrosoft icrosoft A): This is a load balancing feature that allowsfeature that balancing load a is This System firmware can create a table a createfirmwareSystem can I t writes Windows events for server writes server Windowsfort events E S provides a unified view of identity identity of view unified a provides MMS E dition uses this table to apply to table this uses dition M E dition include: dition dition include: dition etadirectory Services (MMS) Services etadirectory D NUM esktop for Administrationfor esktop

A . A

irectory is compatible with compatible is Directory

NUM

Enterprise A topology of the of topology A S, and helps and MMS,

E

dition NUM A 319

Compendium Expanded Physical Memory Space: On 32-bit Intel platforms, Windows Server 2.2.1 Linux 2003, Datacenter Edition supports Physical Address Extension (PAE), which extends system memory capability to 64 GB. On 64-bit Intel platforms, the Linux is a freely available multi-tasking and multi-user operating system. Linux was memory support increases to a maximum of 16 terabytes. invented by Linus Torvalds and developed further by a number of other developers throughout the world. From the outset, Linux was placed under General Public Windows Sockets: Direct access for SANs: This feature enables Windows Sockets License (GPL). The system can be distributed, used and expanded free of charge. applications that use TCP/IP to obtain the performance benefits of storage area In this way, developers have access to all the source codes, thus being able to networks (SANs) without making application modifications. The fundamental integrate new functions easily or to find and eliminate programming bugs quickly. component of this technology is a multilayer Windows Socket service that Thereby drivers for new adapters (SCSI controllers, graphics cards, etc.) can be emulates TCP/IP via native SAN services. integrated very efficiently.

2.1.4 Windows Small Business Server 2003 Presently, Linux is successfully being used by several millions of users worldwide. The user groups vary from private users, training companies, universities, research Windows Small Business Server 2003 is a server solution offering small and centres right through to commercial users and companies, who consider Linux to medium-sized companies features such as e-mail, secure Internet connectivity, be a real alternative to other operating systems. business Intranets, remote access, support for portable devices, as well as an The extensive network support of Linux, including different servers such as application platform for collaboration. Windows Small Business Server 2003 is so Appletalk, Netware or LAN Manager servers as well as the multitude of supported user-friendly that it carries out the network configuration and server installation network protocols, make Linux a secure and reliable network server system. virtually automatically. Windows Small Business Server 2003 is a fourth generation release. With the Microsoft Windows Server 2003 operating system at its core, There are two different ways of obtaining Linux: All the necessary components it incorporates Microsoft Exchange Server 2003 and Microsoft Windows® can be downloaded free of charge from the Internet. This means that an SharePoint Services. Using this combination of technology, and by incorporating individual operating system can be assembled almost free of charge. An innovative management tools, small and medium-sized companies can create a alternative is to use a so-called Distribution, offered by various companies. They faster and more efficient business environment. include a wide range of applications and installation programmes that significantly Windows Small Business Server 2003 comes in two versions. Windows Small simplify the installation of Linux. Business Server 2003 Standard Edition includes: The distributions differ especially in terms of the included components, such as - Windows Server 2003 programming environments, network software and graphical user interfaces. We - Exchange Server 2003 recommend distributions from SuSE or Red Hat. Both these Linux distributions - Outlook 2003 are very sophisticated and include a wide range of documentation as well as a - Windows SharePoint Services graphically supported installation. All transtec Linux systems are certified and offered with the current versions of SuSE and Red Hat. The Premium Edition additionally includes: - FrontPage 2003 In addition to their distributions for PCs and workstations, both SuSE and Red Hat - SQL Server 2000 SP3 offer special packages for server operation. With SuSE these are the SuSE Linux - ISA Server Enterprise Server and the SuSE Linux Standard Server. Apart from the composition of the package specifically for server operation, it is distinguished from the The main features of Windows Small Business Servers 2003 are outlined below. „normal“ distribution by the following points. For one thing, SuSE carries out extensive tests to ensure that the individual packages are compatible with one E-mail, networking and Internet connectivity another and with important business applications. What is more, SuSE guarantees Windows Small Business Server 2003 gives a company everything needed for up to 2 years‘ support for the package, even after the respective version has been efficient use of the Internet. It provides a system that is easy to manage for joint superseded. Equally important for the business environment is the provision of Internet access, a firewall to protect the local network, Internet e-mail based on important patches and updates. Exchange Server and productivity tools such as Microsoft Outlook® Web Access and Remote Web Workplace. SuSE LINUX Enterprise Server The SuSE LINUX Enterprise Server is one of the leading server operating systems 2.2 Unix Operating Systems for professional use in heterogeneous IT environments of all sizes and orientations. It is available for all relevant hardware platforms: for the 32 and 64-bit processors Unix is still the leading operating system in the workstation world. In fact, it is a from AMD and Intel, as well as for the entire eServer series from IBM, including family of operating systems because practically all system manufacturer supply mainframes. their own version of Unix, which, at least as far as the user interface is concerned, The SuSE LINUX Enterprise Server 8 can be used for all company-relevant server differs considerably from the other versions. However, there is a tendency to and network services: from file, print, Web and security, through to application overcome this wide variance of interfaces, as several manufacturers have begun to and middleware solutions, particularly for servers with high availability, data and port their system to alien architectures. network services of critical corporate importance. A single server operating system The Unix implementations can be categorised under two standards: Berkeley Unix with a uniform Linux code basis! (BSD) and AT&T‘s System V Release 4 (SVR4). At present, the SVR4 is ahead of its Advantages: rival - new versions of Unix follow its standard. As a general rule, if a programme - Huge savings in costs is written for one of these two standards, it can be ported to another system of - Enhanced investment security through system maintenance and open source the same standard without major difficulty. code Different standards are also employed for the user interfaces (GUI - Graphical User - More scalability and stability Interface). However, the more recent ones all follow the X11 definition. For several - High performance cluster years, the MOTIF definition - which is also based on that of X11 - has clearly been - High availability cluster progressing. More and more Unix implementations are using this interface, while the use of the competitor interfaces, like OPENLOOK, have been on the decline. SuSE LINUX Standard Server The SuSE LINUX Standard Server 8, which allows all listed services t o be

320 transtec = ttec in the Netherlands and Belgium platform with a large number of software components integrated as standard.integrated as components software of number large a with platform comprehensiveservice a environmentbecomessystem operating 9 Solaris The 9 Solaris with systems all for essential is 8 Solaris software. 8 Solaris the with compatible are use in already are which Web-I A system. the down shutting without systemsoperating of alterations dynamic forpatching hot and uptime; improvingreconfigurationfor dynamical automatic providers); service for (ideal operationscomputer balancing for functions accounting and projects Jobs, abilities: mainframe additional offers also 8 Solaris of synchronisation Java the applications, of software developmentWeb-centric the for 2SE Java includes This the to desktop the fromtechnologies latest the supports also system operating 8 Solaris supported. areTheprocesses simultaneous million a than more and Bytes) 18 of power the to 10 means (that memory main of Exabytes of amount unlimited environments.are other as Windows or restricted longer no is 8 technology. 64-bit Solaris in generation second the is version 8 Solaris The 8 Solaris 2.2.2 SunOS/Solaris applications.many for basis dependable and cost-effective a provides it increase,where the on is well as systemscommercial in Linux of use the trend: general the reflects packages these of developmentThe firewalls.or servers mail e- as such applications, special for packages completeoffers Furthermore, SuSE applications. client/server of operation powerful a of establishment the forsystem operating server versatile a with equipped are departments and organisations server.application an small Thus,as deployed being of capable is 8 Server I the of utilisation efficient and securefor services server numerous from Apart VPN. via datacompany to accesssecure with providedare members staff external way automatically.graphically.entirelyup set are accounts Windows domain network of allocation 8. automatic The Server Standard L INUX SuSE to connected quickly and easily be can hosts Client organisations.small for 8 Server (DNS). an in as Windows environmentsor server print a controller,Windowsas domain and use instance,forfile wizards: configuration graphical of way by user-friendly is services server the of configurationThe hosts.remote from evena Webvia interface quickly and easily modified and configured be can that (ACL)AccessControlLists by management permission detailed using by enhanced (LD service directory central a in administered is improved.substantially and revised fully been have rights user and file of administration and assignment The costs.provider saving and filter and Webbuffersecurity contents, increasingtoconfigured be the interface, graphical a of means By name.company/user the of information address,the network the is 8 Server Standard L INUX SuSE of management and setup for needed now is that All program.maintenance SuSE the CPU two to up supports and A foravailable is Server Standardservices. The of implementation secure and reliable seeking departments and organisations small at graphically,configureddirected is irectory Access Directory 2 Java A nternet and file/print services under Windows and services file/print and nternet

nstall Tool makes the software installation very simple. Solaris applications Solaris simple. very nstall installation Tool software the makes I X Standard LINUX SuSE integratedin also is server e-mail full a addition, n

P M

latform, edia Framework for media streaming, X-Server video extensions, the extensions, video streaming,X-Server Frameworkmedia foredia

P Enterprise rotocol(L D PD

I As (e.g.As nternet addresses (more than a 1 with 38 zeros), 18 zeros), 38 with 1 a than (more addresses nternet

s. The package includes support and maintenance bymaintenance and support includes s. package The Ultra E ) directory server provide the foundation forfoundation the provide server directory AP) P ) application server and a Lightweight a and server application EE) (J2 dition alm

or III P I ilot) and ilot) X Standard Server 8, user data user 8, Server Standard L INUX SuSE n I ) can be configured be can (DHCP) addresses IP

Ultra nternet access, e-mail, print and file and print access,e-mail, nternet I ntranet and a reliable basis for the for basis reliable a and ntranet e processor.IIe etworked Real Real processing.Time Networked and MD I nternet gateway and name server server name gatewayand nternet ). Security of data has been has data of Security AP). X Standard LINUX SuSE Unix, I ntel 32-bit processors (x86) processors 32-bit ntel I nternet provider and the and providernternet I nternet access can access nternet transtec = ttec in the the in ttec = transtec P ractically an ractically I nternet. I n this n N etherlands and Belgium and etherlands I the and e-mails via spread only systems.operating nowadays practically are These and platformscomputer all nearly on viruses computerare there fact: sad A RA in disk, hard disk, floppy on programmestored a or file,executable an itself.copyingbe by area can This memory a executed,changes if which, sequence instruction an is virus computer A 2.3 Computer Viruses efficiency.increased and savings cost to lead all productivity increased downtime,and unplanned reduced customers’resolution, problem simplified errors, of ratereduced platforms.A Sun’s RAS 3.0 Cluster Sun the integrationwith operations.enhanced of include costThese total the lower to designed are Solaris new the featuresin Numerous oads. l server typical forpercent 10 to up by performance systemincreases algorithm colouring page novel A benchmark). SWIM accordingto (measurement times three almost performance HPC increasingpages, memory largefor support with features. security key of series a by security added decisive the provides Solaris of version fully and standard as supplied solutions security by services network forrequired security of level high the delivers 9 Solaris 9. Solaris with working Volume purchases. product additional no with consolidation server out-of-the-box Resource 9 Solaris efficiencies. management system bringing tasks,routine in time valuable staff Software Flash Solaris the as featuressuch with minutes 20 takes operation hour 4- formerly a was expansions.system Whatsmooth and fast featuressupport management change new with operations of cost the lower helps also 9 Solaris solutions. individual popular with comparison invitingapproach an system,operating core the in servers LD and application integratedJ2EE fully include environmenttosystem Sun the integrationof full include will updatesComing costs.incremental no with demand, on services Server Application the With provisioning.and security management, systems,volume file as well as servers, integrating directory Web,and application computing.of ways the infrastructure. integrationthe with The of applications key for costs acquisition reducedsignificantly the through operations of cost total the lower helps 9 Solaris Volume Solaris as systemstandard the of part a much as just are and management identity The servers. high-end for 5-40% by performance systemincrease to optimized localization, The with test internal (Sun improves library multithreading Enhanced improvements: performance followingfeatures the deliveredhavenew the of redesigning. Some or recoding recompiling,without faster run applications make to able areprocessors Cu III the and 9 environments. Solaris 64-bit scalable and reliable highly mandate (HPC) computing performance high and enterprise of demands The cost.incremental no at end-to-end secure applications all make to systemoperating core the of part all featuresare security Sun the via accesssecure Kerberosserver, and V5 access.remote encrypted products. availableseparately nternet. Sun UltraSP UltraSP

M M Open SunScreen 3.2 SunScreen Sun P anager Resource anager,9 Solaris rofile technology and processes provide the best solutions to optimize to solutions best the provideprocesses and technology rofile ARC ARC ARC ONE Net software and configuration management features.management configuration and software processor has a has processor III processor III contributes to high data availability and maximum reliability in reliability maximum and availability data high to contributes included, Solaris 9 makes it easier to deploy scalable, reliablescalable, deploy to easier it makes 9 Solaris included, irectory Server Directory Environment(Sun M D anager . Complete and accurate configuration control savescontrolaccurateconfiguration and Complete . elivering services through the Sun the through services elivering Application Server, making Solaris 9 the only operating only the 9 Solaris Server,making Application ONE Oracle

ships with a firewall which need fear no comparison with comparison no fear need which firewall a with ships O brings integrated resource management that allows thatintegrated resourcemanagement brings , etc., or also in RA in also or etc., M, ther security features of Solaris 9 include Secure LD Secure include 9 Solaris featuresof security ther

on-chip memory controller memory on-chip

xpress) by up to four times.four to up by Express) Solaris Secure Shell Secure Solaris

D fully integrated and the integratedand fully ual Translation(D Buffer Lookaside M ONE) anager and a multitude of further products. further of multitude a and anager integrated security services integratedsecurity

brings new efficiencies and new and efficiencies new brings

nline Analytical Online M. provides functions for secure,for functions provides ONE

improveavailability P allows data and code and data allows ortal Server. robustortal These platform means platform ONE provisioning

Sun P rocessing ( rocessing ONE UltraSP I n addition, n . new The and TLB) directory directory AP OLAP) ARC ARC and IT

Solaris Solaris 321

AP ,

Compendium Types of Viruses System viruses use particular programme components, like the master boot record, partition boot record, diskette boot sector, FAT, and the operating system’s File Viruses root directory as the infection medium. These system areas are connected to the File viruses attach themselves to selected programmes, employing various respective data media, and infection occurs when the respective data media is techniques. They are spread whenever an already infected programme is called up. used. Resident file viruses represent an intensified version in as much as they take root in the memory after they are called up. Thus, each time a programme is executed, Boot sector viruses attack exclusively the boot sector on floppy disk or hard infection is repeated. This type of virus is harder to eliminate; in some cases, disks. The infection spreads slowly, since it can only take place during booting up it can even survive a warm start of the computer. Most of the viruses known from an infected floppy disk. today belong to this category. So-called stealth viruses were developed to go transtec recommends that systems should be continuously checked with the help undetected by virus scanners, checksum programmes, or protection programmes. of suitable virus software in order to prevent extensive damages. Similar to the U.S. Stealth bomber, they possess a camouflage mechanism that enables them to conceal their existence. Polymorph viruses are the latest and most dangerous kind of file viruses. This category consists of viruses, which alter their appearance for each new infection. Normally, they do this by re-encoding the virus code with a key which varies with each infection.

322 transtec = ttec in the Netherlands and Belgium 3. Cluster C Cluster HPC an of Setup Ideal paid to the interconnect in the parallel mode. parallel the in interconnect the to paid be to has attention special network, the via place takes this Sincesynchronised. result. end I in. come results the as soon as sentorder. is correct task the in next Thethem arrangesnecessary, if and, check plausibility quick a after them archives which end,front the to back sent are results processing,the After nodes. the to values start or jobs various sends nodes.end frontthe The all on programruns same mode. parallel and batch both in used be can cluster A by met frequently are nodes the integratedin locally disk hard the of speed the on made demands required.not is The swapping that such selected be should memory main the of the between particularly made be should comparison variants,processor For the application. specific a for examined cluster.be the should differentarchitecturesForof optimisation, performance the determine type network the as well as nodes of type and number The 5. ID RA in disks hard multiple over distributed data the and 1 ID RA in mirrored a via connected unit supply powerredundant a have should failure.it whycluster totalis Thatto leads failureFront end nodes. the of status operating the monitor and receiptfollowingresults the archiveto jobs, with nodes the supply to queue, a in users the of jobs the administer to is performs end front the task The thousands-mark. the reach means all by can but 256, and 8 between practice in is sector.computernodes personal of number The C inexpensive rule a As end.front the to results partial the send and work actual the do They network. a via reached be to nodes compute the administers end frontThe network. dedicated a via processors the to distributed then are system.jobs operating computing Thethe on superposed resources.middleware,arethe called to programs,applicationalso These its distributing interface software a of consists cluster a viewpoint, user‘sFrom the administration. simple and scalability good the arealternative an as cluster a of favour in factors advantage,price the scarcer.fromincreasingly becoming Apart aredesign-related reasonsfor used be to has mainframe a which in - sectors The result a at faster arrive and time saveto wants - user The accuracy higher with calculated result the haveto wants - user The D U share. market appreciable an conquered ratio,however, on performance clusters HPC encountered.classically are solutions I High 3.1 system on the cluster. Clusters are typically operated with Linux. with operated cluster. typically the are Clusterson system operating the of configuration and set-up the in chosen be can Variouscourses n a parallel mode, the compute nodes operate simultaneously on a common a on simultaneously operate nodes compute the mode, parallel a n (High sector mainframe the n se of a mainframe is generally considered for three reasons: threefor considered generally is mainframe a of se ue to its complexity, the problem is in principle unsolvable on single systems single on unsolvable principle in is problemcomplexity, the its to ue I n this case the results have to be matched, the data has to be to has data the matched, be tohave results the case this n P erformance Computing Clusters Computing erformance O TS (common-off-the-shelf) hardware is used, as is known from the from known is as used, hardwareis (common-off-the-shelf) TS hard disks.hard IDE P C) diverse mainframe diverse Computing,HPC) erformance

I ntel Xeon and A and Xeonntel D ue to their vastly improvedprice/ vastly their to ue

A32 basis have in recent yearsrecent in have basis IA32

S, the boot hard disk should be should disk hard boot the UPS,

MD I pteron CPU Opteron n the former case the case former the n

transtec = ttec in the the in ttec = transtec U se of Windowsse

. size The

N etherlands and Belgium and etherlands The speedup, hence the performance gained, depends on the degree to which todegree the on depends gained, performance the speedup,hence The pointless. usually is mode parallel the in performance increasingfor expansion I basis. case-to-case a on resolved be to has for go to technology which of question protocol. proprietary The a of use the by achieved is hardware,this optimised or Quadrics SCI long.transfer,too data often of is beginning the until time technology,coordinationthis the however,segment,of latency performance the the in than higher significantly still are switch GigE a of costs the only standard, as nodes the in included usually Gigabit diversified. is this forcards network of range mode. parallel Thethe in area memory adjacent an in data the to access rapid achieve to order in necessary is network high-performance A nodes. computing the in distributed locally memories has cluster HPC an system,multi-processor) (symmetric SMP as memory sharedcentral a has which mainframe, a Unlike results.intermediate of storage external for only serves consequently disk card.hard network local Thethe via booting during loaded are which end,front the on deposited are images respective the Hereconcept. boot-from-LA N the accordingto installation an is alternativeThe nodes. all on update kernel individual requires systemoperating the Forof maintenanceinstance,solution. this with drawback a be can overheadsnodes.administration increasedthe The on and end front the on both systemoperating the of installation local the is version classical The administrator‘spreferences.the on depends distribution Linux the of knowledge. selection special The demands and uncommon is group which is managed and controlled by a cluster software. cluster a bycontrolled and managed is which group A Cluster High-Availability 3.2 transtec.of whitepaper HPC the in found be can informationFurther future.in widespread more become to sure are technology,clusters sophisticated HPC the and solution the of sequences. film making or business insurance and banking in mining dataresearch, genetic and drug calculations,current simulation, test crash uses. of Typicalare examplesvariety wide a to put meanwhileare clusters HPC performance. uniform of subclusters I system. the in node slowest the of speed effective the with only but practice, loss. integratedwithout is performance extra native the mode,batch a in used system performance. a Withsame the havetohave necessarily not do nodes Additional difficulty.any without out carried be also can extension Similarly,subsequent himself.nodes defective change easily can administrator costs.upfront the reducesThe required, which not therefore generally is nodes. of number total the to proportion in performance critical. not usually is node computing Failurea of system. conditioning air an with together underestimated.be not should CPU/U 4 to density packing the increasesystems Blade facilitated. is maintenanceimproved and be can nodes the of cooling CPU/U 2 to up to enclosures standard with CPU/U 0.5 from increases density advisable. packing is Thecluster HPC 19-system a of set-up The node. single a of speed I 9. factor the has speedup the node, single a of speed the times nine achieves parallelised. is application the n practice therefore the administration will in this case be more likely to define to likely more be case this in will administrationtherefore the practice n the times 48 only achieve would application same the used were nodes 100 f t is helpful to compare experimental results in a test environment.test a in results experimental compareto helpful is t high- availability high- I n a parallel mode, a faster node will not operate with full performance in performance full with operate not will node faster a mode, parallel a n I nfiniBand networks are up to factor 10 faster here. Apart from an from here.Apart faster 10 factor to up are networks nfiniBand cluster refers to the linkage of several systems to form a form to systemsseveral of linkage the refersto cluster

I f, for example, when 10 nodes are used the cluster the used are nodes 10 when f,example,for bit/s range. For parallel operation in the high- the in operationFor range.parallel Mbit/s D ependable operation is often possible only possible often is operation ependable

. The waste heat of a cluster a of heat . waste The I t reduces the cluster the reduces t E thernet is meanwhile is thernet

n-site service service On-site . Furthermore,the . D P D ue to the stability stability the to ue hysically,cluster a olphin, xtreme node Extreme M yrinet, yrinet, 323

Compendium is a group of two or more independent servers that serve the same group of only occasionally. The applications with several threads can run as pseudo-parallel clients and can access the same data. Putting this into practice with the latest processes on one server with one processor or as genuine parallel processes in technologies means, in general, that servers are connected via the usual I/O symmetrical multi-processor systems (SMP). buses and a standard network for client access. From a logical point of view, a In a cluster, groups of the application threads can be executed on different servers cluster represents a single administrative unit, in which any server can provide any because they all have access to the same data. When an application becomes too authorised client with any available service. The servers must have access to the large for one server, a second server can be installed, thus forming a cluster and same data and be equipped with a common security system. According to the enhancing the application capacity. latest technological standards, this means that the servers connected to a cluster Permitting servers to access the same data requires co-ordination. This co- generally have the same architecture and run with the same version of an identical ordination can be accomplished by a database manager or a distributed data operating system. system. In a cluster with no co-ordination, simultaneous and direct access by Although there are various possibilities for the set-up of clusters, they all provide several servers to any data file is not possible. However, some applications can be the following three advantages: scaled even with this limited accessibility. The applications can be distributed so - Fault-tolerant high availability for applications and data that the individual programmes use different data. - Scalability of hardware resources - Simpler administration of large or fast-growing systems. Simpler Administration The third advantage of clusters is simpler administration. Clusters simplify the Higher Availability complexity of system administration by enhancing the scope of the applications, In company-critical environments, the high availability of services (e.g. web servers, data and user domains administered by a single system. Among other areas, databases, or network file systems) is the basis for success of a company. Common system administration comprises the following: reasons for problems with the availability of services are the different forms of - Operating system system malfunctions. A malfunction can be due to the hardware, the software, the - Middleware application or the system execution. Protection against these malfunctions can be - Application maintenance provided using a cluster as an application server or a data server. Besides providing - Administration of user accounts redundancy for computing performance, I/O and storage components, clusters - Configuration management and also enable one server to function for another server without interruption, if the - Data backup. latter breaks down. In this way, clusters allow high availability of server services The complexity and cost of system administration depend on the size and even when malfunctions occur. especially the number of the included systems. For instance, running daily data The illustration depicts a simple cluster, which consists of two networked servers backups is necessary for all servers that store important data, independent of the and a RAID system for . amount of data to be secured. In addition, user-account modifications must be updated on all servers accessed by the user. Clusters reduce the number of individual systems and thereby also the cost of system administration by integrating a large number of applications, data and users into one computer system. One of the advantages is that a cluster system must contain only one set each of user accounts, data access authorisations, data backup rules, applications, data base managers, etc. Although the individual systems may vary due to the different cluster architectures used, it is normally more efficient to administer one cluster rather than the corresponding number of unconnected server systems.

The load distribution between the individual cluster systems is carried out automatically. Both Unix and OpenVMS clusters feature automatic load distribution and scalability, whereas Windows 2000 clusters provide greater availability only. With the Cluster Server of the Windows 2000 Advanced Server, two servers can be connected within a cluster, with the Windows 2000 Datacenter Server, four servers can be connected. The automatic IP load distribution, which is independent from the Cluster server, allows the distribution of Web applications on up to 32 systems. Microsoft Cluster Server Configuration

- When hardware or software malfunctions occur, or when a server fails, the other High Availability Cluster with Linux FailSafe server can take over its functions The Linux FailSafe cluster software is a universal, freely configurable, and error- - When the network interface of a server fails so that the connection to the clients tolerant high-availability system, with which individual services can be redundantly is interrupted, the clients still have the possibility of using the second server set up. This software enables a service to be automatically or manually migrated - When an I/O bus or an I/O adapter fails, the data of the RAID system can be to another node (server) in case of an error. An error is not necessarily due to accessed via an alternative path defective hardware: application errors, desolate processes, and critical system - When a disk fails, the data can still be accessed via the RAID system. conditions are also recognised as such and handled accordingly. In most cases the In all cases, both the servers and the clients must identify and deal with the individual nodes of the cluster must be able to access common data sections in malfunction. It is true that the cluster service performance is interrupted, but order to take over the data in case of an error. An SCSI RAID or SAN architecture with clusters, this interruption normally lasts only a few seconds and not several can be used here, which allows the servers to take over data sections in case of minutes or hours as is the case with conventional restoration methods. an error. If the data sections were successfully taken over, then the active cluster server can restart the services and make them available using the same IP. Thus, Scalability the downtime of a service can be calculated and even controlled (within certain Another advantage of some cluster architectures is their scalability, which permits limitations), since various criteria can be configured for the migration of services. applications to be expanded beyond the capacity of a single server. Many With the corresponding redundant server systems, an availability of 99.99% can applications have several threads of relatively delimited activities, which interact be achieved with this solution.

324 transtec = ttec in the Netherlands and Belgium 4. Storage buses transmission media are defined. are media transmission physical various the which in level interface the Finally,is levellowest the interfaces. other of suppliers for easier access make Storage P the uses SCI) 16-bit and (8-bit another. interface SCI classical The one with communicatedevices individual the how and interface, respective the on appears command SCI a protocolshowcontrol interface. logical Theseof types various the protocolsfor the concerns which protocollevel the is level command this Below commands). these of up made is level command so-called (the Streamdisks, hard for devices: of groups individual for commands primary these on based commands special are C SCI called are what by formed is levels.SCI of coreinterface The and protocolcommand, of consisting rules, of system three-layer a incorporates SCI (American ANSI by 1986 in passed was standard other.each SCI with communicate first can The adapter host the and drives the which overbus, peripheral a but interface, classical a actually not is SCI since themselves,drives the in intelligence of deal great a incorporatedevices SCI configuration. easy and flexible and rate transmission maximum relativelyhigh a were SCI of benefits the outset, Fromthe widespread. very become has and systemscomputer various to peripherals other and scanners for occasionally and types, all of drives storage mass of connection the for used is which defined, serially later and parallel,wide, (Small SCI 4.2 SCI Fibre channelwas intended busfrom asaserial theword go. SCI willbecome SCI. serial storage buseswilltherefore ATA allbeserial: willbecome ATA serial (S-ATA) and entailing that somepresent-day busesare serial faster thanparallel ones. Future more difficult. Thisis resultinginadeadendthedevelopment ofparallel buses, reaching itslimits: The synchronisation oftheparallel data signals isbecoming ever secured by increasing theclockrate. However, intransmission speedis thisstep-up more intelligent protocols andenhanced transmission reliability, suchwere then from by acquiring newstandards harnessing Apart withhigherperformance bandwidth from theoutset. bits. There were alsonon-realised planswith32bits. IDEwas operated with8-bit 32 timesover. Thus theSCIbuswas initiallydefinedwith8bits, later with16 I bandwidth: P 4.1.1 P adapters. ofstorageconnection by products transfer to thesebuseswiththeaidofATA peripheral buses, are for usedprimarily workstations andnotebooks, alsofor the I high performance. the workstation andFibre sector Channel,CSI protocol, likewiseusingtheS for the cost-effective solution for internal storage systems, in theSCIbusprimarily SSA mainlyfrom IBM),three standards have establishedthemselves: ATA (IDE)as today are ofnooronlyminorsignificance (e.g.: IPI,DSSIinthedigital environment, and thelike). Whereas there buses, were which ofproprietary initiallyaseries of hard disks, tapedrives andopticalstorage systems to thesystem bus(PCI,SBus Storage busesare understood to theconnection beperipheral busespermitting 4.1 I technologies: Fibretechnologies: serial new of introduction the in manifested is modularity this of advantage The n comparison with serial connections,n comparison withserial thethroughput increases 8,16oreven arallel busesinitiallypromised to provide ahighpotential ofadditional rotocol(IPC ommands (SPC ommands ntroduction arallel versus buses serial P C ) for SSA as well as the Generic Generic the as well as SSA for rotocol(SSP) ontroller nterface nterface C ). The new interfaces are based on their own protocols: the Serial Serial protocols:the own their on based are interfaces ). new The omputer Systemsomputer ). All devices must be able to process these commands. these processThereto able be must devices All ). C hannel and SSA can be integrated without difficulty, in what, difficulty,in integratedwithout be can SSA and Channel n addition,FireWire andUSB(niversalBus),thegeneral Serial ommands (SCCommands C

ommands (SCCommands

I nterface) is an is nterface)

) for host adapters, adapters, host for ) ) for tape drives, tape for )

bus in its original form 8-bit form original its in bus I/O ational Standards National

P acketized M ultimedia C

transtec = ttec in the the in ttec = transtec ommands (SBC ommands P ) to rotocol(GPP) I C nterlocked I nstitute). ommands P rimary rimary ) N etherlands and Belgium and etherlands - Fibre ChannelArbitrated LoopC-AL) (F - Fibre Channel - SSA SCI Serial - Ultra320 SCI - Ultra160 SCI - Ultra2 SCI - Ultra SCI - SCI-2 - SCI-1 P P interfaces. protocolsand various the detail more in describes section followingThe (5 phase commandslower a and SCI) Ultra320 (320 phase data faster a between SCI parallel with imposed division the and achieved,areratestransfer data higher substantially SSA), Channel, (Fibre interfaces serial downwardcompatible.the is using SCI-3 When SCI-3. called is transmission serial to 2) and (1 SCI parallel original the of extension The interfaces. new introducing of process the up other,speeds the this on and expensesaves this SCI. namely standard,widespread most the is doubt, without -2 specifications (including Fastand (including SCI specifications SC SI-2 complete the support automatically 20 to up of rates transferdata provides Fastand SCI Wideof S CSI SCI). combination The (Wide bits 16 to width path data the increasing of option the as well as bus), 8-bit 10 of ratetransfer maximum higher a provides ( set command standardised and tape disks,hard for improvedsupport with set command expanded featuresan SCI-2 improvements.significantoffers and 1986 in developed was standard SC SI-2 The SCI-2 5 mode synchronous in MB/s, 1 max. of ratesreached mode asynchronous in run percent.data of transferThe approximately90 is transfer data for market. overheadtoday's The in drives. tape and S CSI-1. was passed standard SCI first The SCI-1 4.2.1 P P * 3mfor 4drives arallel SCI arallel SCI arallel SCIinterfaces

arallel SCI drives, MO nterfaces

I B/s on the bus. the on MB/s ts other performance standards have rendered it uncompetitive it renderedhavestandards performance other ts CD -R s, scanners and jukeboxes.and scanners OMs, C ommon -1 has no longer been used. been longer no has SCI-1 1990, Since MB/s. ot all devices that claim to support SCI-2 support to claim that devices all Not C -2 specification SCI-2 the Set), ommand I t supported primarily only hard disks hard only primarily supported t , with the standard the (Fastwith MB/s SCI, B/s) is no longer needed. longer no is MB/s) I n addition to the new the to addition n n the one hand, one the On B/s with MB/s

325

compendium Wide SCSI) but generally, most of the newer drives use Fast SCSI. Some of the connect 8-bit devices to a 16-bit bus; however, a number of special configuration Wide SCSI-2 devices (16-bit data bus) have been misnamed SCSI-3 and non-Fast rules must be followed. SCSI-2 devices have been called SCSI-1-compliant, which can be misleading. Single-ended and differential or LVD SCSI cannot be operated on the bus simultaneously. Attempting to do so can result in damage to the drive and Ultra SCSI controller. Only LVD drives adjust independently to single-ended when an SE drive The transfer clock rate was doubled again on the Ultra SCSI, a fully downward or SE controller is found on the bus. compatible extension of SCSI-2. By doubling the clock rate on the bus, the transfer rate is also doubled to 20 MB/s (8 bit) or 40 MB/s (16 bit). However, Synchronous and Asynchronous Data Transfer this only applies to the data transfer rate. Command transmission continues to With parallel SCSI, data can be transferred in either asynchronous or the faster take place at 5 MB/s. Plugs and cables were also kept for compatibility reasons. synchronous mode. When data is transferred asynchronously, each byte is sent However, this increase in performance does have its price: the maximum and confirmed separately; whereas in synchronous mode, multiple bytes are sent permissible cable length is shortened to 1.5 metres in the case of single-ended and then confirmed together as a whole. Thus, the overhead is lower and the and 12.5 metres in the case of differential cabling. An active termination is transfer rate higher in synchronous mode. Generally, all peripheral devices are essential in order to ensure interference- free transmission. able to work in asynchronous mode. Synchronous drives and controllers perform handshaking before data transfer, to determine whether or not the partner device To simplify the task of configuring the SCSI bus for the user, the Ultra SCSI is capable of transferring data synchronously. The best transfer mode is then specification contains the SCAM (SCSI Configuration AutoMatically) feature. In automatically selected. Current SCSI drives and host adapters always support the other words, the connected device configures itself. The implementation of SCAM synchronous transfer mode. in the devices, however, is left up to each manufacturer. Cable for Parallel SCSI Ultra2 SCSI (LVD - Low Voltage Differential) A few basic factors should be considered when selecting the proper SCSI cable to The products which incorporated Ultra2 SCSI went on the market in 1998. ensure correct and error-free data transfer: Once again the data transfer rate was doubled to 80 MB/s at 16 bit. Since SCSI cables must comply with UL (Underwriter Laboratories) and CSA (Canadian further halving of the cable length for single-ended cabling would make the Standard Association) standards. Individual cable strands should be made of bus practically unusable, this standard was only defined as LVD (Low Voltage copper braid, or better, tinned copper. They should be twisted in pairs, and Differential) as a differential cabling which only works with one third of the signal then up to one meter of the entire cable bundle should be twisted once more. voltage. The maximum cable length is still as much as 12 m. The entire cable should then be double shielded, which is usually done with a silver foil and an additional wire mesh over the cable bundle. If multiple devices To keep this standard compatible with the previous ones, Ultra2 drives are are to be connected to the SCSI bus, the individual cables should be kept as equipped with an auto-sensing interface, which enables operation with short as possible, and optimally be all of the same length. This reduces the bus’s conventional single-ended controllers (but not differential), although only at Ultra susceptibility to interference. SCSI speeds. However, they are not equipped with internal termination, which means that separate terminators must be used, even inside the computer. Since Wide SCSI transmits over a 16-bit bus instead of an 8-bit bus, the 50-pin SCSI cable is not sufficient. Wide SCSI, therefore, uses 68-pin cables. Ultra160 SCSI Due to the fact that the cables for single-ended and differential SCSI are identical, Ultra160 uses the same cabling as Ultra2 LVD-SCSI. The doubling of the transfer the same cables can be used for both types of interfaces. rate was achieved by the data transfer on the rising and falling edge of the signal. Terminating the Subsystems Ultra320 SCSI With growing data transfer rates, the demands on cable and SCSI bus termination Products incorporating Ultra320 have been available since 2002. It is however are also increasing. Differential SCSI and active termination are indispensable with highly improbable that the bandwidth will be doubled again with Ultra 640 SCSI, current SCSI buses. as serial standards such as Fibre Channel and serial ATA, later also serial SCSI, will increasingly threaten the survival of parallel SCSI. In contrast to passive terminators, active terminators operate with an integrated Even Ultra320 raises great technical difficulties for imp lementation in external voltage controller. They keep the terminator power line to exactly 2.85 V with storage solutions. active components. With passive termination, they are kept to about 3 V with a passive distributor. Depending on the cable type and length, the voltage in cables SCSI-3 with passive termination can greatly vary. SCSI-3 conforms to the standards of both parallel SCSI and those for serial interfaces described later. The Forced Perfect Terminator (FPT) is a version of active bus termination mainly used in the IBM world (RS/6000). FPT terminators dynamically adjust SCSI bus SCSI Transmission Standards impedance. With FPT, however, this type of termination must be used at both The parallel SCSI interface comes in different versions distinguished by the method ends of the bus including the . of transferring data. SCSI signals are transmitted either on 8-bit (narrow) or 16-bit It is generally recommended to use active terminators, since they offer the SCSI (Wide SCSI) buses. Up to seven drives can be attached to an 8-bit bus, and up to bus better protection against electromagnetic interference compared with passive 15 drives to a 16-bit bus. Both bus widths have the cable types: single-ended (SE) termination. or differential (D), or, with present-day standards, low voltage differential (LVD). SE SCSI signals are transmitted on a single wire, while D SCSI signals are transmitted Connection of Subsystems on two twisted wires. The latter are therefore less susceptible to magnetic There are the following possibilities when connecting external SCSI subsystems: interference. The benefits of D SCSI include improved interference immunity and thus longer cable lengths. 1. Up to now no external subsystem has been connected to the SCSI port: In this The only drawback of D SCSI is that drives, host adapters and termination are case, the appropriate SCSI cable for the computer platform is required. Suitable more expensive than those for SE SCSI. cable exchange options for the corresponding computer platform are outlined in When selecting a subsystem, it is important to take into account that the host the price list section for all external subsystems. Thus instead of standard cables, adapter SCSI interface must match that of the drive. Basically it is possible to cables with the plugs stated in the price list are available. The subsystem can be

326 transtec = ttec in the Netherlands and Belgium increase as soon as their speed and their function become familiar among these among familiar become function their and speed their as soon as increase Fibrefast very workstations.of and use systemsThe desktop some in found be can interfaces fibre-optic system-specificpresent, At inexpensive in use any find well. as workstations in systems,later mainframe in applied mainly Fibre circulation. limited with interfaces physical of number increasing the for Relief - simultaneously) (optionally interface physicalcommon a protocolsvia various of transport the permits this interface; physical the protocolby- logical Tothe transport volumes data large - of Totransmission high-speed enable types differentinterface these - networking Tofor means a offer interfaces physical - of To differenttypes support are:development this of goals principal The directions.several standard the to extension an transfer.as information 1988, in began developmentThis flexible a protocolsfor new establish to ANSI by further developed being Fibre a via cables a via made are connections fibre-optic The implemented. be can systems high-end systems,and low-cost and low-end both interface, same the and cables.one with Thus, fibre-optic or copperrequire rates shielded Twisteda is version P simplest Fibrea in mixed be can recognition. types Variouscable error simple allows which code 10-bit to 8 an for used is ratetransfergross higher 132 with higher little a areratestransfer 12.5 of ratestransfer media, Forboth conductors. copper for also but cables, fibre-optic for only not Fibrename The 4.3 Fibre Channel 2005. or before2004 expected not are rates.First products transmission higher with buses SCI permit future in will hence and below described Channel Fibrethe than cost-effective more standards.be SCI will They serial on out Since Attached4.2.2 Serial SCI(SAS) Terminatorschapter. SCI the in listed are cables required.Suitable is subsystems existing workstation. the for suitable plug a with fitted cable a for cable standard exchangingthe - required.is plug with terminatorcorresponding Furthermore,plug.a suitable the with fitted cable a for cable box.standard exchangingrequiresthe This - options: two are therecase, 3. subsystems.existing the between will at inserted or to added be can subsystem new the cable, supplied 2. etc.). (computers,disks enclosures the in lengths cable including well, as considered be to has always length cable required. permissible is terminator maximum Thea Furthermore,plug. correct the with computer of type each to connected directly obtain data transfer with high-speed performance. performance. high-speed with transferdata obtain to order in networked be tohave workgroupsthat small users. includes This C 68 plugs. HD68 with equipped not are but presentare subsystems External plugs: HD68 with equipped and presentare subsystems External onnecting the new subsystem directly to the workstation. requiresthe Thisto directly subsystem new the onnecting

hannel can be found in systems of varying sizes, but at the moment it is it moment the at but sizes, varying of systems in found be can Channel is and developed was that array standard a for term general the is Channel standard, work is currently being carried carried being currently is workstandard, SCI parallel last the is Ultra320 I ntelligent (High Speed Serial Serial Speed (High HSSDC 9-pin or plug 9-pin DSub I n addition, a further cable for connecting the new subsystem to the to subsystem new the connecting for cable further a addition, n

hannel is somewhat misleading, for the serial bus is specified is bus serial the formisleading, somewhat is Channel

P B/s to approx.400 to MB/s eripheral

PCs. C onnecting the new subsystem to the last existing last the to subsystem new the onnecting I nterface (IPI) nterface

air cable. Large distances or high transfer high or distances Large cable. air bit/s to 2 Gbit/s.slightly resulting2 Theto Mbit/s B/s are intended.are physicalMB/s The D

Enhanced or L or SC uplex

hannel interfaces will interfaces Channel P hannel system. Channel The hysical and branched into branched and hysical plug, the copper the plug, C

transtec = ttec in the the in ttec = transtec

DC ) plug. ) I U t is unlikely to unlikely is t C sing the sing ables and ables

I n this n

N etherlands and Belgium and etherlands Networks importance. secondary of is time delay transmission which error-freedelivery,in is transmission data requirementfor important An data. transmit to used are Channels slaves.the are devices peripheral master,the the and is host Here,the system. host the to directly connected P host. this to connected channels the in knowledge the of all comprises system host The levels. tables.configuration the in or software host the in modificationsrequires also change every and advance in known are host a with communicating of capable devices all which environmentin predictable and structured closed, a in operate channels rule, a As Channels communication. host-to-host a support networks rule, a As participation. software with flow information an to back traces which input processingsystem an refersto an as soon as software of participation no to little hardwarewith the in out carried being transferdata the by possible as low as kept is system.input processing systemperipheral The and host between data of amounts large transports that computer) host a networks. and channels named: be can communicationsystem protocolsfor Two peripheral basic A major contribution of the Fibrethe of contribution major A as such LANs and network) optical synchronous = SONE installed.company includes been Thisalreadyhave which LANs to as well as networks global important most the to basis Channel Fibrea on work which systems connecting for aspects important are These Fibrethat see P the with familiar are who users All traffic. network other and 802 IEEE FibreThe users. network and channel both new a in processescommunication Fibre Supported flicker.not does image the as long as noticed be not will route,this en lost arepixels two or one hand, other the if,on use; no of be will data late,the is role. secondary a playing delivery error-freepriority,with top has delivery here,timely transmission; video for and speech for also but transmission, error-freedata for used only not are Networks system. telephone the to made often are analogies reason,access. Forthis equal havetelephones all operates,wheresystem telephone the which in manner the type. same the of systems other to comparison in access equal havemainframe a and workstationprotocols. A network of help the with informationexchange they time, to time from others; networked. be can mainframes a called access,is environment.”equal “peer-to-peer have and workstationsAny all devices connected the that assumes environment,which unstructured This service. software correct the to transactions routing and sessions transmission up setting authorisation, access checking for support software intensivemore time.demands any situation at This device other every environment. unpredictable basically and unstructured open, an in operate networks hand, other the On eripherals such as magnetic tape drives and disk drives as well as printers are printers as well as drives disk and drives tape magnetic as such eripherals ), e-mail, file transfer,other file and logons remotee-mail, rotocol(IP), O hannel seeks to combine the best aspects of each of these opposing these of each of aspects best the combine to seeks Channel

ccasionally, this is also described as a as described “master-slavealso ccasionally,is environment”.this

hannel supports transmission of ATof transmission (Asynchronous M supports Channel Transfer P interface that usually implies numerous small transfers with larger with transfers small numerous implies usually that interface I/O rotocols

hannel supports these protocols with higher speeds. higher protocolswith these supports Channel U sually, the term channel refers to the peripheral the refersto channel term sually,the

M operation commences. The term network, in contrast, in commences.network, operation term I/O The ost channels can manage these more complex knowledge knowledge complex more these manage can channels ost early every host or device can communicate with communicate can device or host every Nearly I n the process, each system is independent of the of independent is system process,each the n I n this sense, this environment is comparable to comparable environmentis this sense, this n

hannel is that both interface types, that is, that types, interface both that is Channel I f, for instance, delivery of a video transmission video a of f,delivery instance,for interface, which meet the requirementsof the meet which interface, I/O E thernet. thernet.

T-based(SONE systems I nternet services will services nternet

interface (with interface I/O

I nternet

M

ode), 327 T

compendium channels and networks, can now share the same physical medium. Over the past out by the fabric system instead of the ports. two years, I/O channels have been extended to include network applications (e.g. This peer-to-peer service is an essential component of the Fibre Channel design. A with the help of SCSI, to network two workstations). In the same way, with the system designed for peer-to-peer services can be utilised in such a way that the help of network data transfer protocols, networks move data back and forth host-type master-slave communication process is emulated. In this way, Fibre between systems and file servers (e.g. Network File System NFS). Channel is able to simultaneously support channel and network protocols.

With the Fibre Channel, it is now possible to use the same physical medium and Like a Telephone System physical transport protocol by means of a common hardware port in order to The function of a fabric system can be compared to that of a telephone system - manage both channel and network activities. It is possible to send information to we dial a number, the telephone system finds the path to the requested phone a network con nected to a workstation via Fibre Channel and at the same time to number, the phone rings and the person being called answers. If a switch or link use Fibre Channel for in-house communication with local peripherals (e.g. with crashes, the telephone company routes the calls via other paths, which the caller hard disk and tape drives). rarely notices. Most of us are unaware of the intermediate links which the telephone company employs in order to make our simple phone call a success. Fibre Channel protocol features: Fibre Channel does not have an instruction set such as SCSI and IPI, but rather provides a mechanism for placing other protocols However, we do provide the telephone company with some information regarding on Fibre Channel. This is possible because Fibre Channel serves as a carrier for our phone call. For example, the telephone number begins (in the U.S.A.) with the these instruction sets and in this way the recipient can distinguish between the digit “1”, followed by ten digits in the form of an area code (3), switches (3) and two. This implies that various instruction sets for older I/O interfaces, for which the subscriber number (4). If a “1” is not used at the beginning of the telephone software outlay was previously necessary, can be directly applied on Fibre number, the caller is making a local call, and only seven digits will be used. This Channel. information assists the telephone company in setting up the connection. The telephone number corresponds to the Fibre Channel address ID. Part of the Separation of I/O operations from the physical I/O interface is a significant Fibre address ID is used in order to specify the fabric system domain which will be Channel feature, and it enables the simultaneous use of different instruction sets. affected; the rest is used to determine the special port. The various instruction sets, such as SCSI, IPI-3, IP etc., are usually used on their own special interfaces. Fibre Channel, however, defines a single, physical It should be remembered that the telephone system is not involved in the content transmission mechanism for these instruction sets. of the discussion between the two participants in the phone call (or affected by it); it is only responsible for establishing the connection. In the same way, Fibre Fibre Channel Channel is responsible for establishing the link, and the protocols which are - Is not aware of the content or significance of the information which has just placed on it (e.g. SCSI or IPI) carry the instructions. These protocols play a role been transmitted similar to that of languages in telephone systems. Fibre Channel and other - Increases connectivity from dozens to hundreds or even thousands of devices protocols should be seen as integral components in the exchange of information. - Increases the maximum distance between devices - Increases the transmission rate four or fivefold in comparison with the most The fabric system may consist of a single or several fabric elements. Just as with widespread channels and a hundredfold in comparison with the most common the telephone system, we don’t know (or don’t pay attention to the question of) networks. how many switches we must go through as long as we are connected with the correct destination station. The following sections describe how Fibre Channel permits network setup. A fabric system is also called a switched topology or crosspoint switching Networking Topologies topology. Routing by means of various switches takes place by the fabric elements Fibre Channel devices are also called nodes; each has at least one port to permit interpreting the destination ID in the frame as soon as it arrives in each fabric access to the outside world (in other words, to another node). The components, element. which link two or more ports with one another, are together described as the “topology”. All Fibre Channel systems have just these two elements: nodes with The fabric system can be physically implemented as a single fabric element with ports and topologies. multiple F Ports or it can be implemented as a series of multiple interconnected fabric elements. Routing or switching of each connection is transparent for both Each Fibre Channel port uses a wire pair - one wire to transmit information to the N_Ports, which are connected to the fabric outside edge by means of F_Ports. port, and one wire to transmit information leaving the port. Fibre Channel have either electric wires or optical wave guides. This fibre pair is called a link and is a If the topology is disconnected from the nodes, as is the case with the telephone component of every topology. system and Fibre Channel, new technologies can be introduced for the router. New speeds and new functions can be implemented in the fabric system without Data is always transmitted in units (so-called frames) by means of these links. The losing all previous investment in existing nodes. Fibre Channel permits a Fibre Channel standard defines three topologies; however, stress is placed above combination of accessory devices with varying speeds or capabilities. all on a topology which is also known as the fabric system. It is this topology which will be described first. Other Topologies In addition to the fabric system topology, the Fibre Channel standard defines two Fabric Topology other topologies. One is called “point-to-point topology” with just two connected A fabric system permits dynamic connection between nodes by means of the ports. In this case, routing does not take place. The third topology is called ports which are connected to this system. Please keep in mind that this “arbitrated loop”. This is a simple and inexpensive topology for connecting several application of the term “Fabric” may also be used as a synonym for the terms dozen NL_Ports on one loop. The ports in an arbitrated loop topology, which are “Switch” or “Router”. Each port in a node, a so-called N port or NL port, is called NL_Ports and FL_Ports, are slightly different from N_Ports and F_Ports. They connected to the fabric system by means of a link. Each port in a fabric system is contain all the functions of N_Ports and F_Ports and are capable of working called an F port. With the help of the fabric system’s services, every node is able properly in a fabric system. An FL_Port is a port in a fabric system which processes to communicate with every other port connected to other F ports in the same the arbitrated loop protocol. fabric system. With this type of topology, all frame routing operations are carried With arbitrated loop topology, each port sees each message (as in the token ring

328 transtec = ttec in the Netherlands and Belgium D FC-0 FC-x. labelled is each standard; Fibrethe accordingto layers function called areinterests frame. of lost differentfields Thesea of because appeared that problems concerning questions with do to much not have optics fibre- the for connection testable and reliable a establishing with deal that Aspects Fibrefor true holds together,also belong naturallywhich topics Some subscriber’snumber.and ator’soper code,area with number telephone a to comparableare components altogether.bit 24 to comes These which bit), (8 port and bit), (8 region bit), I time. some integratedforsystems individual requirementsfor coverall should address 24-bit a with system fabric one to time same the at logged-on be can that ports million 16 overslightly to frame.amounts the limitation Thisof the of numbers many How system. fabric the within increased has communication of distance the reliability,speed,and since speed, higher to due nodes the to on passed indirectly is technology new the of benefit the change to having without implemented be can itself,technology systemimprovements of fabric the Within available.is system fabric a if distances long or short overcommunication a types media and speeds varying with ports allow systems system. fabric the to connected or ports extension called elements, fabric among connections of types other add to capability The node. the to connected directly not are that and areageographical larger a over out spread are that elements between connection the bymastered be to big too are that nodes between distances bridge SONE like networks node. the torates performance and grades service of variety a add to order in system one in other each with merged Fibrefromtopologies loop arbitrated and system fabric The down. breaks link or element an that event the for paths standby of provisionfacilitates traffic. handle also Thisto order in elements fabric between paths numeroushave can system fabric a service, reliableprovide to order in this does also and points various between routesnumerousconfigure can companytelephone the as Just destinations.invalid for frames rejects system fabric Similarly,unobtainable.the is number this that I destination the to data the routessystem fabric the address,and destination a enter you function: same Fibresystem. telephone the of responsibility the is Routing call. youfriend’s yourwhen reach tohouse takes system telephone the number.telephone friend’sa dial youcomprehensible: easily more be to topology system fabric a of function the fororder in below continued be will analogy system telephone The protocol.acquisition token no have which messages those ignores and over passes protocol)and H norm. norm. FC-PH in written connections. is layerThis port the for featuresnecessary performance optical and electrical as well n the fabric system topology,addresssystem the fabric the n you notify will companynumber,telephonetelephone the incorrect an dial you f efines the physical Fibrephysical the efines

N_P orts can be implemented? The fabric system is solely limited by the by limited solely is systemimplemented? fabric be The can orts N_P E_P T or ATor T SONE over M I orts that are displayed in the target address field in the header the in field address target the in displayedare that orts n order to do so, you do not need to know the exact route exact the know to need not do so,you do to order n

N_P orts, means an increase in value of every disk drive that is that drive disk every of value in increase an means orts, N

orts. These special connections can exist among fabric fabric among exist can connections special orts. These P

hannel elements, including media type, connections as connections type, media including elements, Channel ort. hannel standard. Five layers are defined by this by Fivestandard.defined arelayers Channel

N_P

P articular attributes of Fibreof attributes articular orts in any way. The major proportion of the way.of any in proportion major Theorts I n addition, the fabric system can use other use can system fabric the addition, n T among fabric elements, in order to order in elements, fabric among T is divided into three parts: domain (8 domain parts: three into divided is ID

hannel’s fabric system has the Channel’shas system fabric

hannel can be can Channel hannel and fabric fabric and Channel transtec = ttec in the the in ttec = transtec Channel. ID . This

N etherlands and Belgium and etherlands specifies the operating environment which is necessary for the application (e.g.application the for necessary is environmentwhich operating the specifies that profile a defining on working already aregroups industrial Some chosen. be must manufacturer individual an which from options numerous supports also FC-2s Layer media. and speed of differentcombinations implement can node an in implemented be can option type media services. more or one implement can node Each admits. norm the differentoptions the all on take cannot node a FC-4, layer On interestedin not probably system. the foreigninformationto unnecessary with occupied not is party third a that so norm, extra an contains set command Fibreuse that sets D FC-4 norm. no is therelayer For node.this D FC-3 layer.this writing of predominantly consists norm login. FC-PH user The the protocolslike and protocol-independentframes particular Additionally,defines rules.header, applicationit frame and the of D FC-2 norm. FC-PH in written order,is layerdetection. Thiserror and D FC-1 servers. smaller in greaterrole a playing also meanwhileare they needs storageincreasing to due systems,but ID RA end 1996. since available been have products FC-AL normally.operateto continues bus the and interference. by blocked or use in not is that port any bypasses hub the so bus,entire the blocking circuit, the break will defect cable or failure device loop,a Channel Fibreby used also is principle operation. same The during drive a exchangingof the is which function, hot-plug the support to also and drive the of configurationautomatic the is backplane the of task further A closed. remains circuit the and slot, empty the bypasses logic backplane lines. supply power and data both SC 40-pin a via backplane the to connected are disks hardlink. The cable normal the as well architecture,as I possible. not is definition) Fibregeneral the with (as devices several between exchangedata bus. the of control obtain first transfer,must data device this initiatethe information. the process should it whether checks which port, read the via first device the reaches packet data ring. a as arranged are ports) (NL ports loop 127 of maximum arrays.A disk and devices local for Fibreof type this of topology the describes and LoopArbitrated Fibrethe of version slimmed-down A Fibre different.requirementsare the since design, port a on decides beforehe the install to plans integratedthe thatprovider system A made.are expansions the when available is list request port master a that considered,so be should FC-4 options, expansion additional planning or product a developing When HIPPI, SCI, - AL supports the backplane backplane the supports AL better,arraysFC- disc the handle to able be to order n efines the mapping between the lower Fibrelower the between mapping the efines one in differentports at available be can that services used commonly efines layout,contentsframe the protocol,including framing and signalling the efines transfercoding,word 8-bit/10-bit the protocol,transferincluding the efines

hannel Arbitrated Loop (FC-AL) LoopArbitrated Channel -3 etc.). IPI-3 C onsequently, the flow of data to other devices is not interrupted not is devices other to data of onsequently,flow the hannel. Here you can find SCI, find can you Here Channel. I ata exchange is only possible as a point-to-point link. link. point-to-point a as possible only is exchangeData nternet -1 there are no options. no are there FC-1 layer On A (Single A

-3 or HIPPI. or IPI-3 P ) later should carefully consider both FC-4s both consider carefully should later rotocol(IP) I f not, it resends it via the write port. port. write the via it resends it not, f I C f there is no drive in one of the bays,the the of one in drive no is there f onnector Attachment) plug, which includes which plug,Attachment) onnector

hannel is the Fibrethe is Channel

port. port. N hannel layers and the command the and layers Channel I I -3 you are you SCI-3 with work you f -0, solely a speed and speed a solely FC-0, layer On nitially they were limited to high- to limitedwere they nitially hannel hubs. Channel , and SBCS. and HIPPI, IPI-3, ach individual port in a in port individual Each

hannel AL. AL means AL AL. Channel -3 protocoland IPI-3

hannel designed Channel I Channel n a Fibrea n I n order to order n P

arallel Each Each 329

compendium 4.4 SSA For Enhanced IDE, there are ATAPI (AT Attachment Packed Interface) extensions, which - as for SCSI - allow peripherals like CD-ROM, tape drives, scanners etc. to Serial Storage Architecture (SSA) is a high-performance interface that combines be connected. I/O devices of all platforms. This serial bus-like interface was developed by IBM based on IBM-9333 UltraATA technology, and is predominantly used for connecting hard disks. With this, up to Much like Ultra SCSI, UltraATA increases the clock rate on the bus of the Fast ATA 128 devices can be connected together. interface. Therefore with UltraATA a data transfer rate of 33 MB/s can be achieved on the bus. Like the Fibre Channel, SSA is a point-to-point connection, but two write and two read channels are available. On each channel (read/write), a maximum transfer UltraATA/66, UltraATA/100, UltraATA/133 rate of 20 MB/s is possible, which corresponds to a cumulated transfer rate of 80 After three further increases, the clock rate has meanwhile reached 133 MHz. MB/s. However, this rate can only be achieved if the read/write ratio is 1:1 and the Cabling requirement has become more stringent since ATA100 due to the host adapter accesses data located on at least four disks. increased clock rate: A special 80-pin cable is needed. However, because the connector remains unchanged, older disks and controllers can still be combined A standard TwistedPair cable is sufficient for distances of up to 20 m between with the new standards, of course at the cost of reduced performance. two devices to be connected. Fibre-optic cables are used in lengths of up to 680 m. Only 6 % of the data carried over SSA is used for control or routing functions, S-ATA i.e. 94 % of the data is user data. However, only a few manufacturers (like IBM) ATA technologies have gained a surprising lead over SCSI during the past few support SSA, and it has meanwhile been almost entirely superseded by Fibre years, driven mainly by the significantly cheaper prices, and the new standard Channel Arbitrated Loop technology. will contribute to strengthening this trend. Again, it is a serial technology, rather than a parallel, that is gaining in popularity: serial ATA (S-ATA) provides a bus that 4.5 ATA (IDE) Interface offers a high growth potential for the transfer rate. It can also manage a spectrum of features from the SCSI world such as command queueing. S-ATA thereby The Integrated Drive Electronics (IDE) interface is very widespread in the PC becomes usable for applications with high I/O demands - and thus also provides environment. In fact, it has now completely replaced the older ST506 interfaces. high performance in RAID systems. Up to two hard disks can be connected to each IDE interface. If two IDE hard disks are to be operated, one disk is configured as the master, and the other disk as the 4.6 USB Interface slave. Note that as standard, IDE hard disks are configured as a master drive and not as a slave. USB (Universal Serial Bus) is an interface standard for connecting external devices The original IDE interface, also called AT Bus or ATA (AT Attachment), could (for example, printers, modems, keyboards, monitors, digital cameras, etc.) to PCs theoretically transfer a maximum of 4.3 MB/s. However, in the real world, only via a bus system. The advantage of a USB interface is the trouble-free integration 2 MB/s were reached. It was used solely for the connection of hard disks and of additional periphery. The USB controller recognises if additional periphery has supports no other peripherals. A number of disk manufacturers have improved the been connected, and automatically installs the required drivers and prepares them IDE interface in order to meet the increased performance requirements. for operation. Complicated new configurations of the PC systems are a thing of the past. Theoretically, up to 127 devices can be connected to the USB interface.

Fast ATA and Enhanced IDE The transition to USB version 2.0 achieved a higher transfer rate of 400 MBit/s The Integrated Drive Electronics (IDE) interface is still widespread in the PC and some extensions of the protocol. environment. It originates from the ST506 interface. Up to two hard disks can be Detailed information on USB can be found at www..org. connected to each IDE interface. If two IDE hard disks are to be operated, one disk is configured as the master, and the other disk as the slave. Note that as 4.7 FireWire 1394 standard, IDE hard disks are configured as a master drive and not as a slave. The original IDE interface, also called AT Bus or ATA (AT Attachment), could FireWire (IEEE 1394) is a high-speed, serial data bus with maximal 800 Mbit/s theoretically transfer a maximum of 4.3 MB/s. However, in the real world, only which, originally used by Apple, is used in particular to connect multimedia about 2 MB/s were reached. It was intended solely for the connection of hard disk devices such as camcorders and digital cameras to the computer. The standard drives and hence supported no other peripherals. The IDE interface was developed was defined in 1995. The bus enables reserved bandwidths, which is particularly further by several disk manufacturers in order to meet the increased performance required in the area of video. 16 devices can be connected in series over a length requirements. of 72 m (4.5 m per link). However, there are also bridges which increase the number of devices drastically by cascading (a maximum of 1023 bridges are Fast ATA and Enhanced IDE permitted). The differences between these two developments were primarily in the marketing FireWire uses a 6-pin special cable that is made from two shielded TwistedPair strategy, while the implementations and functions were for the most part wires and two additional wires for the power supply. identical. Both Fast ATA and Enhanced IDE remained compatible with the old IDE FireWire is currently of relatively great significance in the notebook sector and in adapters and hard disks. They used the same 40-pin cable. the Apple environment.

330 transtec = ttec in the Netherlands and Belgium 5. Hard whole, there can only be one active partition, the active partition usually being usually partition active the partition, active one be only can therewhole, partition. primary one only with configured be can disk Each C year.following the of needs the include to enough large is that recommended is disk hard enough. a Thus long is upgrade the and purchase initial the between time the if only but disk, larger a investingin initially than economical more be proveto could disk larger P needed. initially is than more much purchasing not however,forprices,reason a is disk hard in challenged.decrease are constant Thelimits capacity the when suffers assumed. originally than needed usually is space more as planned generously be should capacity disk hard The annually.double to expected currently is requiredspace disk data. and The applications with up fill quickly can large,quite seemed originally that disk a result, a Asspace. disk of lot a programmesrequireapplication and data. of volumes high particularly handle toprogramshave publishing programsand graphicapplications, (A/V) video Audio/ application. the on depends storage mass of amountrequired The parameters. certain within vary can formatting.the upon dependent also is capacity net The entered.is command df the when user the to this display not does storage,and auxiliary as system file a of 10% additional an reserves Solaris example,(for data administrationfor space some requiressystem file Furthermore,the characters). or 1 calculatesystems operating most 1 calculate manufacturers a of definition the on agreed not havesystems operating and disks hard of manufacturers that is reason Another information.address and correction error for blocks the between used is capacity the of percent 20 to 10 because user. the topartly available is Thisactually capacity net of amount the fromconsiderably differs capacity net the as manufacturer the by listed figure The available. actually is that space storage of amount total the is capacity net surfaces. The disk of number the and tracks of number the track‘slength, the by track a of density bit the multiplying by calculateddrive, the of capacity theoretical maximum the refersto capacities. capacity grossnet The and gross M 5.2.1 C time.access and speed rotational as featuressuch performance are factor.these to Added form and interface differentiatedbus capacity,arestorageaccordingto disks Hard necessary. as often as to written and from read be can recordingprocedureand magnetic applications.a standard use most Theyfor used devices storage mass the are disks) magnetic or as Winchesterdisks known also (originally disks Hard 5.2 Hard Diskrives of Arrays (Redundant IDs RA form to combined are disks tolerance,hard fault increased ensure to also I drives. disk floppy and disks e.g.drives,hard disk hard magnetic and memories semiconductor includes blocks). category informationThis to access (direct access random by available data makes which devices storage here.mass memories are dataThese primary so-called about Wetalking are 5.1 I size 8 GB. 8 size disk maximum the GB, 2 was partition a of size maximum Thepartition. boot the n order to manage large data volumes exceeding the capacity of a hard disk and disk hard a of capacity exceedingthe volumes data large manage to order n urchasing a smaller disk with little reserve capacity and later upgrading to a to upgradinglater and capacity reserve little with disk smaller a urchasing apacity restrictions also existed under existed also restrictions apacity ass storage capacity is measured in gigabytes, with a differentiationbetween a gigabytes,with in measured is capacity storage ass ntroduction apacity apacity P lease note, however, that the block size with a 2 GB partition is 32 is partition GB 2 a with however,sizenote, block lease the that D isks and RA I ndependent B = 1,000 x 1,000 Byte (= 1,000,000 Bytes), while Bytes), 1,000,000 (= Byte 1,000 x 1,000 = MB

ne should also consider that the performance performance the that consider also should One Disks). B = 1,024 x 1,024 Byte (= 1,048,576 Bytes 1,048,576 (= Byte 1,024 x 1,024 = MB

DOS, Win3.1, and Win3.11,Windows95. I n addition, the operating systemoperating the addition, n ID

I t is not absolute, but absolute, not is t

I transtec = ttec in the the in ttec = transtec n the system, as a as system, the n M egabyte.

Disk N etherlands and Belgium and etherlands 5.2.4 P GB. approx.300 of capacity net a have and popular most the probably are heights.versions mm 3.5“ The 12.5 and mm 19 in and drives, 1“) = mm (25.4 profile low and mm) (41 height half offeredas are disks available.Hard also are factors form 1.3“ and 1.8“ with C (electronics). logicsmotor, the spindle and the actuator), (the heads the moving for mechanics the heads,read/write the arebox the inside Also box). drive the inside disk (the medium a of diameter the refersto factor form The 5.2.3 Form Factor chapter.Buses Storage externally.used ATAWhile FC (rarely SCI or ATA,with producedcurrently are disks HardFibre and SCI 5.2.2 Storage BusI configured. 16 is system file 500 a using when case the automatically is which KB, 8 of size block a configuring by avoided be only KB. can 32 This of block a occupies still file 250-byte a i.e.KB, Though 512 bytes/sector is most common, disks can be formatted with as many as with formatted be can disks common, most is bytes/sector 512 Though rate.transfer the influences also sector) a intogrouped are bytes many how (i.e. size Block etc.read, be can cylinder complete a positioned, already is head the i.e.conditions, optimal under calculated are figures these that note should one e.g.period, latency the during tasks other process cannot that systemsoperating for true especially performance. is systemThis the better the time, total the shorter The adapter. host the to bus the via sent controller,then disk is hard it the at arrived has data transfer.data the forrequired is time transfer less rate,transferthe disk‘srate.transfera the higher into The it convert will 8 by frequency the given. is frequency a Sometimes second. transferredper bytes of number important. becomesrate transferdisk‘s (internal) the stage this At controller.disk hard the to disk hard the transferredfrom actually is data point, this Atms. 4.17 of latency a have disks rpm 7200 So,disk‘sspeed.rotational the upon exclusivelydependent specified.alwaysis figure is revolution This a of 1/2 forrequiredtime. wait time rotationalThe or latency called is time wait This reached. is block desired the until wait must it track, correct the on positioned is head read/write time. settling the includes already time seek extinguished. the drives,With be to oscillations for movementor head a after rest mechanically to order in requires mechanism head the time much how specifies time settling disk. The the across distance maximum the of travel1/3 to mechanism head the forrequired time the i.e.time, access mean the always is time.given valueseek The called is time surface. This disk the on point specific a at heads read/write the position to drive the overhead.systemcauses operating request the This called is process controller.disk hard this the forrequiredto time passed The is which request a generatessystem operating the data, write or read to programwants a When processes: sequential multiple forrequired time the from calculated rate.transferis data time accessthe The and data the access torequired time the by influenced primarily is time. computer‘sresponse performance the The for factor important most the is storage mass primary the of performance the memory, computer‘smain the of size the and speed processor the from Apart urrently, hard disks are available in 2.5“ and 3.5“ form factors. Smaller disks Smaller factors. form 3.5“ and 2.5“ in availableare urrently, disks hard B partition. With the current Windows versions the maximum size of the of size currentmaximum the the WindowsWithversionspartition. MB erformance Featureserformance M ore on bus systems can be found in the Storage Buses chapter.Buses Storage the in found be can systems bus on ore MS

S. When comparing transfer rates of hard disks in tables, in disks hard of ratestransfer comparingDOS. When M xabytes (16 million Terabytes). A 20 GB partition can easily be easily can million Terabytes).(16 partition Exabytes GB 20 A

nterfaces nterfaces ore on the differences of the interfaces is to be found in the in found be to is interfaces the differencesof the on ore are FC or SCI computers,only in internally used is ) hannel interface. interface. Channel

I nce the Once t indicates the indicates t

Dividing I f the f 331

compendium as 4,096 bytes/sector. Larger sectors leave more space for user data, because less disk uses special software to allocate RAM from main memory, which is then space is required for storing formatting information. This consequently leads to organised as a virtual disk. A solid state disk is a hardware solution which consists a higher transfer rate. This is, however, not possible with all operating systems, of a bus interface and memory modules. Both types of memory provide disk-like nor is it appropriate for all applications. The smaller the read/write files are, the access. As with a real disk, files can be stored and subsequently read. smaller the sector size should be (e.g. databases). The larger the files, the more appropriate large sectors are (e.g. for graphics processing). Advantage: The seek time and rotational wait time do not apply; thus, data is What is more, for the drives the transfer rates are often indicated through the transferred with the maximum possible speed. bus. But since the drives have a built-in cache, this transfer rate only reflects the path from the disk cache to the host adapter cache. The actual transfer rate from Disadvantage: These types of storage are not suitable for permanent data storage the disk surface into the host adapter‘s cache is significantly smaller. (even with a battery backup). They are also considerably more expensive than disk drives of the same capacity. A battery backup is always required for this type Most applications only read and write a few blocks, thus generating low-volume of storage, as even the shortest power loss could result in total data loss. This transfers per request. Therefore, the rate at which the hard disk transfers data to solution has not established itself in practice. the host adapter is not of primary importance. The more segmented the data on a hard disk, the shorter the read access required. One should ensure, however, that 5.4 Floppy Disk Drives the transfer rates of the components in the data path match each other; hard disk - hard disk controller - host controller - computer RAM. Disk technology is based on a removable, rotating magnetic disk with an access Another means of increasing performance is by using controllers with cache time of 150-250 msec and has the advantage of being widely used for the memory (on the hard disk as well as on the SCSI host adapter). exchange with other systems. Its limited capacity of 360 KB to 2.88 MB is a A multi-segment cache is especially advantageous. The cache is dynamically disadvantage. It is no longer sufficient now that powerful PCs have become administered and divided into several areas. Data is stored in the cache, common not only in companies, but also for private use. depending on which area of the disk it came from. Accessing data in the cache This disadvantage has been largely removed by the ZIP drive from IOMEGA. At eliminates delays caused by mechanical components. 750 MB per storage medium, the capacity of this drive is sufficient even for smaller to medium-size software packages. This drive also features an attractive These controllers process write requests like non-caching controllers: each write price/performance ratio. block is transferred to the connected disk. Additionally, the controller notes the However, the small distance that the magnetic head maintains from the storage block‘s address and contents in its internal memory, the cache. The advantage disk can lead to a head crash. This type of contact usually results in the loss of lies in the reduced number of hard disk access. Caching hard disk controllers are the data stored at the location of the crash. A head crash can also occur in the recommended for the following situations: computers with minimal or barely removable disk drives described below. sufficient RAM, with hard disk drives which process a large number of read/write requests per unit of time, and disks using small numbers of blocks per request 5.5 Removable Disk Drives e.g. system disks, for databases, and in general for applications which often alternate reading and writing to the same files. Caching controllers can reduce the Large amounts of data can be managed with removable disks, without losing time average seek time to under a millisecond. in transferring data from one medium to another. In addition, such subsystems are well suited for high-security situations, for example when data needs to be stored The different types of cache are read, read-ahead and write-back cache. A read in a safe overnight. access to a read cache causes the cache memory to be searched for the requested The use of removable disks also makes sense when the location where data is block. If it is found, it is immediately transferred. There is no delay for disk used is constantly changing, as in the case of a programmer, for example, who is positioning and rotation. If the block is not found in the cache, it is retrieved from active at several locations and always needs access to his project data. Since this the disk. method continues to store data on a disk drive, where the danger of data loss remains, an additional data backup on another medium should be provided for. When a read-ahead cache reads a requested block, it also reads the subsequent Whereas removable disk systems were produced as independent products in the few blocks and stores them all in cache memory. A write-back cache stores blocks past (Iomega, Syquest), today standard hard disks (often with 2.5“ form factor) that need to be written to the disk until there are no more competing read or in appropriate canisters are used, since the transport safety has significantly write requests. This has the effect of speeding up write requests for the user. One improved through new technologies, particularly with the influence of notebooks. danger however, is that if there should be a power failure, all of the data stored in cache would be lost. This is not an issue with a read cache, but it could destroy 5.6 RAID the file structure of a system using a write cache. Thus, it is imperative that systems using a write cache also have an uninterr CPU and system memory power roughly doubles every year. In comparison, it uptible power supply (UPS). takes about two years to double the capacity of mass storage devices and seven years to reduce the average seek times by one half. Conventional disk technology Hit rates of over 90 percent can be achieved using a disk cache, depending on evolves at a markedly slower pace than CPU technology. the size of the cache buffer, the cache logic (i.e. which blocks are overwritten Therefore, techniques are needed which enable the more slowly evolving disk when the cache is full) and the type of application. This provides an enormous technology to keep pace with the processing speed and large data amounts of improvement in throughput. current systems - by means of increased access speed and increased capacity. Typical MTBF times (MTBF = Mean Time Between Failure) of drives today are over Many operating systems automatically reserve a part of the main memory for 500,000 hours. When using 100 disk drives with a single host, a failure is likely to cache, such as Windows 2000/XP, NetWare, Unix and OS/2. More main memory occur once every six months. Even if the level of disk reliability were to continue can radically improve the total system performance. to improve over the next few years, this situation would still be unacceptable. Therefore, systems that are fault tolerant are required. 5.3 Semiconductor Disk Storage (RAM Storage) RAID is the generic term used to describe technologies where several disk drives If disk drives cannot deliver the data for a given application quickly enough, a are grouped together into larger systems, often with a higher fault tolerance. semiconductor disk storage (RAM disk or solid state disk) can be used. The RAM RAID stands for Redundant Array of Independent Disks, where „inexpensive“ is

332 transtec = ttec in the Netherlands and Belgium of the physical blocks to the standard length. This means, however,length. means,standard This the prior to that blocks physical the of length the restricting byavoided be can loss capacity a Such result). will blocks 128-byte only then example,fordrives, four and blocks 512-byte (assuming reduced is disk the on blocks the of length physical the if headers, block for room added the by consumed is capacity additional blocks): (e.g.single requests write short for performance its in lies technology this of disadvantage The byte).(approximately one small very are blocks these requests,if read/write of length the considerably,increasedof independentbe can throughput the set, 0 ID RA a in time same approximatelythe at drives the of each transferredto is data Sincedrives. individual the to written blocks the of size the indicates factor drive.logical striping larger The one form to togethergrouped are drives smaller several which in process a striping,for stands ID RA of type This ID Level5.6.1.1 RA 0 costs). hence (and redundancy of degree the and speed read speed, write are: selection application. the on depends levelappropriate an of selection the performance. or misleading: is term the But level. ID security.RA and called throughputare Theydata maximum achieve to disks hardseveral on data distributing differentof waysareThere ID level5.6.1 RA data. used infrequentlyfor changer) automatic with disks (e.g.optical (e.g. device storage mass large relativelyfast, a data, accessedfrequently formirrored) (possibly device storage mass fast very a data, used continuallyfor ID RA FC with accessed.rarely only is data of majority the accessed,while frequently most the is data of number relativelysmall a rule, factor.a As important an is distribution data of below,optimisation described the types ID RA the of configurations.one using When conventionalsingle-disk using than faster it is problems,nor throughput disk all to answer the not is ID RA differentlengths. of requestsfor speed processing of terms in others the fromdiffers below discussed concepts ID RA the recording.data technical/scientific computing,and super applications, graphic of characteristic is second the whereasapplications, multi-user commercial and systemsprocessing transaction applications, database in typical is request of type blocks). first hundredseveralThe (usually requests long and blocks) few a only or one (typically requests short of speed processing the differencesin ID RA with case the not consideration). under is Thisquestion the fordisregarded be will performance increase which techniques other buffers,and read-ahead(caches, request write I transfers.data individual forrequired average time P techniques: storage ID RA tostrategy drive single-disk a from moving when application, the with significantly varies performance the since for called are disks accessing on remarks levels,however,differentID RA basic fewthe intoa deeper much going Before term. appropriatemore a be would 0. level ID RA upon build „Type“ actually not does 1 level ID RA because misleading today.discussed being are term TheID RA a of Different levels actually is „level“ unit). logical one as togethergrouped drives of number a across evenly distributed are blocks (data striping and controllers), separate two using disks two to writes system (one duplexing simultaneously), disks two to data identical system‘swritescontroller (a single mirroring as such processescommon uses ID RA data. of loss a to even or operation of disruption a to or system whole the of failure the to lead doesn‘t drive single a = redundantof failureas referredthe to is if tolerant“solution storage„fault- A of place in read often „independent“. to be modified, must be read. be must modified, be to block physical the of parts contain which disks, the of all block, a writing to n a single-disk drive system, a read request takes just as long to process as a as processto long as just takes request read a system, drive single-disk a n rimary factors include the proportion of read vs. write requests, as well as the as well requests,as write vs.read of proportion the include factors rimary ID RA IDE I level signifies higher redundancy redundancy higher signifies level ID RA higher a that case the not is t ) for less frequently accessed data, and an archiving storage systemstorage archiving an and data, accessedfrequently less for ) E level has its own advantages and disadvantages and disadvantages and advantages own its has level ID RA very

, which also differs from level to level.to also level arefromTherediffers also which , I t is only after this partial information has been has information partial this after only is t

U seful solutions are a hierarchical systemhierarchical a are solutions seful

transtec = ttec in the the in ttec = transtec C riteria for riteria

ach of Each N etherlands and Belgium and etherlands to the individual disk drives are processed sequentially instead of in parallel. in of instead sequentially processedare drives disk individual the to channel.requests data the Thus,single a via disks access usually controllers the as realised, not is transferdata however,higher solution, controllerof advantage the controller.disk the the forintegratedon opting be evenWhen can and systems operating of number a for solution software a as available is striping Host-based security). data increased (no ID RA constitute not does (striping) fails. drive single a if accessed be can data I disks. differenton simultaneously requests short severalprocess to possible is it addition, drives. disk individual of that to comparable is transfers short for performance I evident. more even becomes 0 ID RA of performance inferior the drive, single larger) (correspondingly a with than ratherparallel, in requests short process can which drives, smaller multiple with comparedWhen poorly.rather performs combination requests,however, requests.Forshort long this writing and readingfor rates U disks. respective their to back written be can blocks affected all that substituted ID Level5.6.1.2 RA 1 just one additional parity disk, whose nth bit contains the parity bit for the x nth x the for bit parity the contains bit nth whose disk, parity additional one just is there method, striping the using to written are which disks,hard x to addition data. the backup to parity uses 3 ID RA - 0 ID RA of case the in giveneven not is and - capacity disk double requires 0+1 and 1 ID RA foroperation fail-safe While ID Level5.6.1.4 RA 3 high. very are costs the used, are disks many as twice Since 1: ID RA of case the in as here applies argument same 0. ID RA of performance the with combined is mirroring of security high the mirrored: is capacity total the of Half striping. and mirroring combines 10, ID RA called sometimes 0+1, ID RA ID Level5.6.1.3 RA 0+1 systems.operating various for solution software M however,is, 1 ID RA straightforward.of implementation The data. redundantby up taken are disks the of rest storage,the data original for used is disk hard one of capacity the only high: are 1 However,ID RA for costs the independently.reading sets) two (or drives disk two overspread be now I channels. on distributed are disks the and calculated not is process write redundant the for load processor additional the assuming 0, ID RA or disk individual an of that to equal is available.performance write The Advantage:set. ID RA a in disks other all to written is disk a on data the all that means which mirroring signifies 1 ID RA n both cases, however, there is still the disadvantage that no part of the stored the of part no that disadvantage however,the cases, still both is n there read/writeresulting the selected, is blocks) (several factor striping high a f n the best case, the read performance may be doubled since read requests can requests read since doubled be may performance read the case, best the n 0 provides the advantage of high transfer high of advantage the provides 0 ID RA with factor striping low a sing irroring (of individual disks) is available either as a controller solution or as a as or solution controller a as either available is disks) individual (of irroring

I solution, but the but solution, ID RA fastest the is this cases, most n

I n case of disk failure the data remain data the failure disk of case n I 0 word,ID RA the of sense strict the n

I 333 I n n

compendium failures occur independently of each other. In the real world external influences (such as or a power surge) may result in a much higher probability of failure for several drives at once; therefore, MTBF calculations for RAID systems are to some extent unrealistic. Even if the time between failures of a RAID system is very high, backups must not be forgotten. RAID does not protect against accidental deletions or software bits of the remaining hard disks. If a hard disk fails, the content of the defective errors that destroy data. Neither will worms, Trojan horses, viruses, or bombs disk can be reconstructed bit by bit from this parity disk together with the data cease to exist or develop respect for RAID. of the remaining disks. However, writing data to a disk always requires the parity disk to be changed - both blocks must first be read, and the new parity value is 5.6.2 Criteria of RAID Selection calculated from them and the new data. Therefore, performance is lower than in the case of RAID 0 or 1. RAID systems serve to improve data availability, not performance. They are generally inferior to single drives in terms of speed. However, the use of RAID 5.6.1.5 RAID Level 5 levels 0 through 5 does make sense for storing large data sets which are accessed either rarely or only with long transfers. RAID offers a cost-effective solution for configuring fail-safe mass storage systems. For frequently accessed data, the use of (mirrored) single disks is recommended; for data that is constantly accessed however, the use of RAM disks is still recommended, which can be mirrored with standard Winchester disks, assuming that the number of write requests is not too high. Mirroring should generally be performed in the controller or host adapter due to a lower computer load. If you require a high transfer rate for long transfers, With RAID -5, the parity disk must be accessed for every write request, creating striping should allow several data paths to the computer and therefore be host- a bottle-neck at the parity disk. RAID 5 spreads parity and data blocks across based. all of the drives. creating a bottle-neck at the parity disk. RAID 5 spreads parity and data blocks across all of the drives. Thus each drive is the parity drive for a In any event a RAID (and the computer) should be connected to a USP, because certain block segment. Read requests can be processed faster because they can be otherwise in case of power failure the actual aim, absolute data availability, is no distributed over even more drives. longer achieved. However, RAID 5 is also far inferior to a single drive in terms of access per Megabyte per time unit for short transfers. 5.6.3 RAID Implementation

5.6.1.6 Other RAID Levels Hard disks can be connected to a RAID in three different ways: by RAID software,

by dedicated RAID controllers incorporated in the computer, or as external RAID The aforementioned RAID levels were first described in a study by the University systems regarded by standard controllers as normal, even though uncommonly of California at Berkeley and have since become quasi standards. In addition, a large, hard disks. number of manufacturer-specific RAID levels are available; these are normally only modifications of the previously described RAID levels.

5.6.3.1 Software RAID So where is RAID implemented? There are two possibilities: either in a controller between the computer and the drives (installed as a card in the computer or in a Windows, Solaris, Linux and many other operating systems already support separate box) or in the host computer itself. The advantage of the first solution is grouping of multiple disk drives to form a RAID system as a standard feature. that RAID processes do not place an additional load on the CPU. The advantage The disk drives simply need to be connected internally or externally with an of the second option is that by establishing several different data paths, a higher SCSI host adapter. Configuration is then carried out through the operating transfer rate can be achieved, since not all of the data must pass through the system. This solution is both economical and simple to implement, but does bottleneck at the RAID controller to the host computer. have a few disadvantages as well. Especially when large amounts of data have A number of manufacturers (of disk drives as well as controllers) provide RAID to be transferred, and when multiple users on a network access the RAID implementations, which differ in the following technical specifications: simultaneously, there are consequences for the performance of the computer. In simple implementations, several drives are connected to a bus and are grouped In this case, the entire processing load must be borne by the CPU of the system together using RAID or a RAID-like configuration. Other implementations allocate (distribution of data on the individual disk drives and calculating of parity). one controller per disk drive, thereby increasing fail safety, as even in the event of . a controller failure, data is still accessible. Drives can be permanently integrated. In any case, however, it is better (even if 5.6.3.2 Internal Hardware RAID with PCI RAID Controller more expensive) to install the drives in such a way that they can be exchanged while the system is up and running, so as not to interrupt data access (hot-swap). In this case, as with external hardware , the RAID controller takes over the When a drive is exchanged, RAID systems allow the data for this drive to be entire system load. As a result, the controllers work with consistent performance, reconstructed during normal operation. regardless of the CPU load. The configuration data for the RAID system is found

on every in the RAID system, thereby making them available even There are also other factors which significantly influence the faily safety of RAID when one disk drive or even the whole controller fails. When the controller or a systems: If all the drives are attached to the same power supply, the probability hard disk is exchanged, the RAID configuration data are read into it and a rebuild of failure (but not necessarily the probability of data loss) is higher than with is initiated. separate power supplies for each drive, or some other redundant power distribution system. To avoid a single point of failure, all cables, controllers, and SCSI RAID controllers are equipped with internal connection sockets and an host connectors should be configured redundantly as well. external connector for SCSI hard disk drives. In this way, the hard disk drives of

the RAID system may be installed in the computer or connected externally. Another word of caution: Beware of the misleading statistical premise that drive

334 transtec = ttec in the Netherlands and Belgium Channel. (ATASATA)Fibreor or SCI via computer the to disks. connected hard are They with operatedinternally aresystems ID RA external of proportion increasing An ID 5.6.4 IDERA platforms. order. reverse in but way same the exactly in happens system this in data of Reading performance. improve writing drastically can cache, write-back a as configured cache, large correspondingly A system. ID RA the forrequiredoperations chosen. level ID RA the on parity,depending computes and drives disk individual the to data distributes burdened.then controller longer ID hardwareRA Theno is bus the which after cache,controller. the ID RA into readthe tois data directly There,all bus FC or SCI the transferredvia is computer the of adapter host the from data system the on load the of regardless performance, Like running. and up still is system the while one new a forexchanged be then can controller data. defective or The time of loss any enclosure,without the in controllersecond the toover switch controller,automaticallydefective will systema the of event controllers.redundant additional installing by achieved be may security data Additionalwell. as security data maximum and lengths cable possible shortest the allowing backplane, a to directly connected are disks hard and controllerThe cable. FC a or cable SCI an using adapter computer‘shost the to connected is system enclosure.separate ID a RA Thein housed are disks hard the and controller the Here,solution. ID RA high-end the present IDs) RA CSI SCI-to-S (or IDs HardwareRA Hardware5.6.3.3 External ID RA s always function with consistent with function alwayscontrollers, IDs hardwareRA ID RA PCI C

ompared with ompared I n the meantime, the meantime, the n

s can be used across used be can controllers,IDs hardwareRA ID RA PCI of the system is not hindered by the computing the by hindered not is system the of CPU

CPU . I n this arrangement, this n transtec = ttec in the the in ttec = transtec I n the n IDE N etherlands and Belgium and etherlands become ever shorter.ever become windows backup where cases in particularly important becomes which argument An libraries. other or tape on than faster significantly course of then is and GB per price: for application further,ideal A Fibreexpensivemore far the on an ID RA common a share servers several if drastically cut be can costs the Here network. storage the in like the and archives servers, e-mail servers, file for demand a databases.foralso used is Thereis capacity I standard). the integratedin are queuing command as such characteristics SCI when only ATA serial of generations first the in either,change not but will (This databases. the not considerable,but meanwhile is throughput data the files large of write and readfor since performance, I system. the of redundancy the by forcompensated is expectancy life low with disks inexpensive of use the was: ID YetRA behind concept the expensive. disks“,very initially weresystems ID RA the because I of Array „Redundant I n Storage AreaStorage n fact of point n t is true that nobody will use will nobody that true is t of talk toof instead disks“ customary „independent became later t „inexpensive ID RA IDE ear-line backup becomes affordable on hard disks through the low price low the through disks hard affordableon becomes backup Near-line

. The databases and other and . databases The as ID RA of definition original the to entirely up live IDs RA IDE etworks based on Fibreon based Networks I nexpensive s in situations demanding maximum demanding situations in IDs RA IDE interface is suggested by the low the by suggested is interface FC with ID RA IDE ID RA Channel Disks“,on emphasis the with „inexpensive“. throughput. I/O I/O -intensive applications then have roomhave then applications-intensive hannel as well, only part of the data the of part only well, as Channel .

, and that can by all means be means all by can that and , disks are too slow for slow too are disks IDE

335

compendium 6. Storage Networks

6.1 Introduction: Storage Centralisation - Hardware-based connection protocol between the interfaces - Flexible cable configuration, copper and fibre-optic, up to 10 km distance. While more and more mainframe solutions were replaced by local servers and workstations in recent years, today a centralisation of the servers and storage The transfer rates of Fibre Channel are currently being developed so that in the connected to them is taking place. The reason for this is the use of ever faster bus near future 4 Gbit and 10 Gbit will be reached. Further information can be found technologies, but also, and principally so, the increasing networking of data. in the Fibre Channel section (4.3).

The centralisation of the storage naturally puts high demands on availability. Since Since the entire exchange of information is handled in this memory network, system failure can bring an entire company to a standstill, it is vital that data paths the SAN can take the load off the LAN. When backup drives are connected to and the memories themselves are designed to be redundant. the SAN and not to the LAN, the load can be relieved even further with the The availability of a system is measured in percent. 99.9% availability signifies an corresponding software. This enables an easier administration of the systems and average failure per annum of 8.8 hours, 99.99% an average failure of 53 minutes higher data transfer rates. A storage area network offers the ideal opportunity to and so on. centralise the entire data.

A significant advantage in setting up central storage networks is the possibility of The easiest way to build a storage area network is using the point-to-point using a storage system (e.g. RAID) by several servers at the same time. The servers connection via Fibre Channel. It connects, for example, a RAID system with a share the capacity of the data memory and thereby reduce the costs both of computer. Only a host adapter (HBA), a cable, a Fibre Channel subsystem (RAID, acquisition and administration. This process is called hardware consolidation. etc.), and a loopback plug as termination (which closes the internal loop) are needed. Due to the simple cable connection, this is substantially easier than SCSI. The centralisation of storage solutions requires the use of suitable network technologies. Instead of direct attached storage (DAS), i.e. storage units directly A larger system is built via switches in ring topology form. This ring topology is connected to the individual server via SCSI or Fibre Channel, technologies are called FC-AL and connects several computers to several memory units. The FC- used also providing data over larger distances. On the one hand this can be the AL (Fibre Channel Arbitrated Loop) enables a distribution of devices, lines, and use of network attached storage (NAS). The file systems are provided directly via user times. Fibre Channel does not necessarily need a fibre-optic cable. A simple network to other servers or the clients. copper cable can be used. The only disadvantage is the reduction in distance that the copper can bridge (approx. 30 m). On the other hand, for higher performance and larger data volumes, preference is often given to a technology which supplies the data to the computer at block SAN Software level - hence like a local disk. This takes place in the storage area networks (SAN), Up to now only SAN hardware has been dealt with. To be able to share the data which either use the fast Fibre Channel, or recently iSCI, if the speed does not between common memory units, special SAN software is needed. This software need to be all that high. Here the SCSI data are transported via the Ethernet, uses the storage networks and enables file sharing on the subsystems. Therefore mostly a Gigabit Ethernet. A driver installed on the computer makes the Ethernet each system, regardless of the operating system, has access to each memory unit card appear like an SCSI controller. at any time and can share the data. Therefore it is referred to as a SAN operating system that is, however, set on the “normal” operating system. Companies such

6.2 Storage Area Network (SAN) as Veritas, Mercury or SANergy offer SAN software which creates and synchronises mirroring, and in case of errors activates the mirrored memory unit, and this is all LAN enables access to local systems which use communication protocols such done within the memory network while the applications are online. The software as TCP/IP, NFS, HTTP. LANs are used for transferring data between various host enables an uninterrupted, dynamic change of the application volume. In addition, systems and user access to systems like workstations, terminals and other devices. the software permits an assignment of memory units or of group memory units LANs handle the flow of information between host system and user. On the and a server. Some SAN software applications support clustering to guarantee other hand, SAN enables the sharing of the flow of information with the entire improved availability of data within the memory network. The memory units environment, i.e. between the computers and the memory units and between the are not assigned to an explicit server when clustering, except for when this has memory units themselves. Thus, all computers connected to the SAN have access been configured so that if the server breaks down, the memory units can still be to the memory units. Similar to LAN networks (Ethernet, FDDI, ATM, and Token accessed or assigned to another server. Ring), SAN also has different topologies to build such a storage area network. It is not absolutely necessary to have a software to build a working SAN network. Storage area networks are special memory networks which link the servers and In this case, only the relevant connections between the computers and the memory systems via wideband networks. It does not really matter, at first, what memory units must be made. Ultimately, the memory units are assigned to the type of operating system is installed on a server. This means that several different computers via the operating system where the integrated standard tools of servers with different memory subsystems can access a network. The SCSI any operating system are by all means sufficient. Additional functions require technology is the best known way in which to build a storage area network in its additional software. simplest form. Due to large limitations, such as the number of devices, distances, terminations etc., Fibre Channel technology is however normally used today 6.3 iSCSI as the communication medium for SANs. The characteristics of Fibre Channel make it very suitable as a medium, since a SAN requires a guaranteed bandwidth Internet-SCSI, in short iSCSI, is a standard which was accepted in autumn 2002, (network speed) and low error rates. which like Fibre Channel transfers data at block level, hence making hard disks and other storage units appear as local SCSI drives. However, Ethernet is used as Fibre Channel offers the following advantages: the transport medium, allowing on the one hand use of existing cabling, which saves time and expense, and on the other hand distances of any magnitude, since - 400 MB/s data throughput (via 2 channels) recourse to the is not excluded. - 16 million devices can be addressed per channel Manufacturers such as Cisco and Adaptec are the forerunners in offering products - Serial connection 8/10-bit encoding incorporating this technology, which in the coming years will replace Fibre

336 transtec = ttec in the Netherlands and Belgium obvious: areoffers technology this benefits The CD/V and drives disk hard as such media storageaccessing for case, this in purpose; single a for constructed is device network server,a print or communication a with storage.Here,as attached network as known technologies more servers. file more and morefor call networks in capacity storage largerfor demand increasing ever The Attached6.4 Network Storage (NAS) libraries. in bridges FC and printers in servers print as way same the in solutions,storage into integrated directly be router.iSCI an called is which network, the to interface the establishes server specialised and dedicated a side,storage the On an Engine), a T with itself,together task this to attends card iSCI dedicated the via data the transmits to then data TCand P/IP SCSI driver,the translatesa which via installed either is iSCI side computer the On consideration. prime the not is performance maximum wherever Channel Cs and workstations. Simpler and more cost-effective is the use of use the is cost-effective more and workstations.Simpler and PCs E thernet accelerator. thernet

P I reviously,involvedpurchasingthis n future these servers will probably will servers these future n

E thernet. thernet. C and workstation and PC r - what is better - a - better is what - Or transtec = ttec in the the in ttec = transtec P/IP (TC OE -R OMs.

O

ffload N etherlands and Belgium and etherlands is available as if it were stored on a under one‘s own operating system.operating one‘s own under server file a on storedwere it if as available is dataview, all of point user‘sFrom platform.a the for typical case each in method the using accessed is data network. The the in server single a on ready held were they if as environments,systemmuch operating defined all on available is data All computers. of example,consisting,forWindows, networks mixedfor ideal also servers storage network protocolsmakes of variety Flexibility: Their - reconfigured.and expanded easily be can server,spacestorage storage one than more using changes: configurationWhen and upgrades Simplified - versions.systemoperating updating when overhead no protocols,is there standardised with work servers storage the Sincemaintenance: Simplified - users. network users.for storage enabling and addresses network setting to limited is system this server,of managementfile a installation: Simplified -

D ue to the fact that the network storage server is actually actually is server storage network the that fact the to ue I t is then immediately available to all authorised authorised all to availableimmediately then is t nix and Unix NetWare

337

Compendium 7. Magnetic Tape Storage

Magnetic tape storage systems are today used mainly for the data backup of written from the inside to the outside, as the inner data area is physically better servers. The low costs per GB for drive and tape media as well as the possibility protected. With data compression, the first generation offers a capacity of 200 GB of safely archiving the tapes over several years are decisive factors in this context. and a data transfer rate of max. 30 MB/s, while the second generation available While, even in the age of DVD, optical storage systems are still too expensive to since 2003 doubles capacity and speed. Both types of cartridges have a free-from- serve as an economical backup medium, data backup on Ultra ATA hard disks is on contact cartridge memory (LTO-CM) with 4 KB, in which manufacturer and user the upswing as a result of dropping prices. information is saved. A back software support is necessary for reading or writing the user information. Magnetic tape drives store data sequentially at block level. Accordingly their forte is writing large information blocks in one continuous data stream. On the 7.1.2 DLT/SDLT Drives other hand, magnetic tapes are rather unsuited for rapid access to specific files or information stored in scattered locations. The DLT (Digital Linear Tape) technology records data, like LTO, in a serpentine-like linear way, so that the tracks lie parallel to the tape edge. 1/2“ magnetic tapes are A more sophisticated backup strategy is therefore gaining ground. Whereas the full written in compact cassettes. DLT4000, DLT7000 and DLT8000, the last three drive data backup at the weekend is still written to magnetic tapes and safely archived, generations, use the same DLTtapeIV tape medium and with compression achieve data backups during the week are stored on IDE RAID systems. This enables rapid up to 40, 70 and 80 GB storage capacity. restoration of individual servers or files on a day-to-day basis, while the tapes still Until now, data was written through two channels located on the read/write provide protection against total breakdown. head, at a density of 62500 bits/inch, and the transfer rate was 3 MB/s. In this format, one data unit consisted of twenty 4 KB blocks, each individual data block 7.1 Linear Tape Recording Method equipped with various error detection procedures (parity, (CRC) and error detection check (EDC)). The last four blocks of each unit consist 7.1.1 LTO Drives of four ECC blocks (ECC, error correction check). In the new storage method, Super DLT, the data is written by four channels located on the read/write head, The LTO (Linear Tape Open) technology was developed by Seagate, HP and IBM. at a density of 86000 bits/inch, which increases the transfer rate to 5 MB/s. In this These manufacturers were also the first ones to offer the appropriate drives. Many storage method, a data unit consists of twenty-four 4 KB blocks, which are also other manufacturers have licensed this technology and also offer the appropriate equipped with error detection procedures. The last five blocks of each unit are ECC drives. Due to the fact that a lot of manufacturers support this technology, LTO has blocks. After every write operation the DLT controller carries out a read test. The quickly established itself on the market. The main part of the LTO recording format written data is compared with the contents of the buffer and corrected if an error is made up of various linear channels which all run in the form of a serpent. The is detected. These safety measures guarantee very high data integrity. data is written on tracks which run the entire length of the tape.

Two standards have been developed: Accelis and Ultrium. However, as yet only An additional advantage of this technology is that wear and tear on the tapes and Ultrium has gone into production. The Accelis format was defined for rapid access. read heads is low. This is achieved through the use of stationary magnetic heads The Accelis format uses two tape rolls and starts accessing data from the middle of and a simple tape threading system. Therefore, DLT drives are capable of up to the tape in order to keep access times down to a minimum. 500,000 tape winding cycles per cartridge. DLTtape IV tapes are capable of more than one million cycles, due to the use of the new technology described above.

1 2

Track Recording with DLT Drives

Track Recording with LTO Drives In 2001, Quantum launched the first SuperDLT drives on the market. They are equipped with a new technology which can reach 220 GB at 22 MB/s (2: This means that the path from the start to the end of the tape is exactly the 1 compression). They require the same size of tapes but with a better coating. same. The advantage of this principle is clearly demonstrated when writing or The systems are read-compatible with DLT8000, DLT7000 and DLT4000, which is reading parts of data, however, it is of no advantage when writing or reading the most important sales argument for SuperDLT over LTO. SDLT320, the second the complete tape. A typical application of this format is with HSM (Hierarchical generation of SuperDLT drives with 320 GB capacity at a transfer rate of 32 MB/s Storage Management), where it is often necessary to read individual files from the (2:1 compression), has been available since mid-2002. tape. With the Accelis format, 256 tracks are divided between two data areas. To downwardly round off the DLT product lines, the year 2000 saw the The first generation will offer a capacity of 50 GB with data compression and a introduction of the DLT VS format. The current generation VS-160, at full 2:1 data transfer rate of a maximum of 40 MB/s. The Ultrium format, in comparison, compression, is capable of writing up to 160 GB of data at a throughput rate of is the optimum medium for high capacity applications and exhibits outstanding 16 MB/s. operational security in an automated or stand-alone environment. The Ultrium medium provides only one tape roll and is ideal for backup, restoring and archiving 7.1.3 ADR Drives programs. The Ultrium format divides 384 tracks between four data areas. The

servo track for head positioning is positioned between the different data areas. The ADR (Advanced Digital Recording) technology was developed by Philips and The tape is written in the form of a serpent with eight parallel tracks. The data is is used in the drives from OnStream. The data is recorded on eight parallel tracks.

338 transtec = ttec in the Netherlands and Belgium metres long. The newD offer an almostdoublenative of36GB. capacity The matching tapes willbe175 P storage. The firstmodelsofthelatest 5thgeneration ofDDSdrives from Hewlett- drives offer special features that goalong way toward error-free ensuring data MB/s. I to 40GB(withcompression and150mtape)withadata transfer rate ofupto 6 tape at adata transfer rate of2MB/s. The standard DDS-4 ensures capacities ofup Drives recording format inDD S-3 are capableofrecording up to 24GB ona125m maximum values prove to beover-optimistic. up to 16GBofdata canbestored. I representstypically afigure achievable withdata compression (2:1). Theoretically, of upto standard. 1MB/susingtheDDS-2 The claimof8GBstorage capacity manufacturers. D The developments inD S greater strain onthetape, aswell asonthedrive‘s head. The is only90°,whileExabyte technology usesacoiling angleof112°,whichputs mm wideandinthedrive itself, thetape‘s coiling anglearound thetapedrum tape. I in theaudiofield. I D 7.2.2 D Track Recording Drives withHelicalScan (D guaranteesmanufacturer aminimumarchive life of10years. latest even generation longterm archiving ofcartridges, isnotaproblem. The is minimal. Dispatch andstoragecartridge are unproblematical. And usingthe to reduce theprice perGBfor themedium,andspace requirement ofa (a maximumof40GBwiththeExabyte M was refined. The large amounts ofdata that canbestored onthesecartridges for devices. useasbackup I Exabyte drives emerged from Video8 technology andwere speciallyimproved technology have already claimedalarge share technology ofthebackup market. D tracks diagonallyonto thetape. These tracks canfeature anextremelyhighdensity. and videosector, whichusethehelicalscanmethod. Rotatingheadsrecord the Very highdata densities are achieved withmagnetic tapedevices from theaudio 7.2.1 Exabyte Drives Recording7.2 HelicalScan M withamaximumdata transfer rateper cartridge ofupto 4MB/s. repositioning ofthetape. ADRdrives offer ofup to 50GB(compressed) acapacity adjust itselfto thedata rate ofthesystem whichtherefore avoids time-consuming have variabledata transfer rates ranging from 0.5to 2MB/s. The drive isableto onthetape. information ADRdrives sections This faulty isalsoableto servo detect position thehead. information isstoredServo between thedata tracks whichisusedto accurately ne significant source of errors is a dirty headdrum.media. Onesignificant source oferrors isadirty The installed torage) format was developed by HPandSony andhasbecome thestandard. ackard andSeagate wareackard introduced in2003.At they thesamespeedasDDS-4, AT evices that have beenadapted to meetdata storage requirements usingthis ( D n comparison to the8mmExabyte tapes, however, D t alsoensures aread DDS-3 DDS-3. and andwrite compatibility with DDS-2 AT Drives igital

1 A udio

4

AT drives canstore upto 8GB(120mtape)at transfer rates 2 t usesthehelicalscanmethodto record data onamagnetic T

ape) technology alsoemerged from digital recording methods 7

5

AT are consequence thedirect ofcompetition between 3 AT72 drives canread andDDS-3 andwrite oldDDS-4 10

8

6 n thecourse ofthis, theirrecording technology 13

11

9

... n practice, however, thesetheoretical ethod 12 AT, Exabyte, AIT) ammoth anda170mtape)help AT tapesare only4 DDS transtec = ttec in the the in ttec = transtec ( D igital D head ata N etherlands and Belgium and etherlands a month. cause mechanicaldamage. Cleaningisrecommended 30GB, orat every leastonce recommended bykits beused. themanufacturer I write headsshouldbecleanedregularly. I D be used. For beartheDDSsealfor example,D thesecartridges We recommend that onlyD libraries makeitpossibleto store data withoutoperator intervention. thelarger onescanholdmorecartridges, than100cartridges. oftape These types tapelibraries. operate theso-called The smallesttapelibraries orchangersholdsix The more expensive random access models usuallyrequire to specialsoftware drives. The simplermodelsonlyallow sequential changeover ofthecartridges. There changersfor are D automatic cartridge additional floppiesorCD generated, whichcanrecover the system situations withoutneeding inemergency to be the helpofappropriate cartridges software, thisenablesboot-capable special feature ofHPD written to by thedrives ofotherD the middleofdata track. This guarantees that even tapesthat have been With theaidof constantly monitors thecondition ofthedrive abackup. andthemediaduring out network-wide diagnostics ofdata mediaproblems. I feature iscalled cleaning tapeisdefective, administrator thenetwork isimmediately notified. This I That, combined withacleaningtape, ensures years drive operation. oftrouble-free cleaner I lifeincreases ofmagnetic theservice tapeanddrive. environmentsreduces working ofusingcleaningtapesinnormal the necessity and fits ina3.5“slot. Anew mechanism, whichsignificantlyfeature istheself-cleaning (Advanced Lossless DataC compression (approx. factor 2.6)have increased by theenhanced ALDC algorithm 260 GB(100uncompressed), whereby thechances ofachievingagood the useofadifferentformat. cartridge iscurrently The maximumcapacity approx. helical scantechnology, likeExabyte, dueto butisnotcompatible withit, inpart The 7.2.3 AITDrives Data RecordingHelical Scan f an inserted tapeisdamaged, too oldornolongerusable,f aninserted however, orifthe n case of serious errors,n caseofserious the internal headcleanerisautomatically activated, the -Eight sealfor Exabyte drives. As withallmagnetic tapedrives, theread and AIT automatically seesto thecleaningofheadsat 15-minute intervals. ( A dvanced TapeAlert time-tracking I ntelligent AT drives isthe -R andisanewfirmware developed by HP ompression). I OMS. AT explicitly manufactured for data storage cartridges T ape) launchedby Sony in1997isbasedon8mm function, theread headisalways function, aligned to AT drive canbereliably manufacturers read. A OBDR n addition,justlikeD t is also important that onlycleaning t isalsoimportant ( AT drives aswell asfor allothertape O ne B ncompatible cleaningtapescan utton n addition, TapeAlert D isaster AT drives, thisdrive AT drives andthe , which carries , whichcarries R ecovery). ecovery). With 339

compendium AME media also making a decisive contribution to reducing the cleaning intervals. more on one cartridge. The disadvantages of the 1/2“ drives are the long backup An interesting novelty is also the use of a memory chip MIC (Memory in Cassette) and restore times and the space required for both the magnetic and the in the cartridges. The size is 16 kBit for 25/65 GB media and 64 kBit for all other tapes. ones. On this chip, backup software can store information, which for conventional technologies is only stored at the beginning of the tape. This makes it possible to 7.4 Software for Magnetic Tape Storage halve the average seek time on a tape, which will then be 27 seconds. Today, all major manufacturers of backup software support the AIT drive with the features 7.4.1 Backup Software set out above. The AIT-3 drive available since summer 2001 has put Sony into the performance Due to the unavoidable possibility of data loss, whether it is caused by human class of LTO and SDLT drives. By using 230 m tapes, the drive achieves a data error, hardware failure, sabotage or viruses, backups cannot be neglected. transfer rate of up to 31 MB/s (compressed) with a capacity of max. 260 GB. Sony Although it is possible to carry out a backup without supporting software, will continue the AIT format at least until the year 2005 through three further ever larger amounts of data and the use of libraries have made this an almost generations reaching the TB range. At special feature is that the drives have read impractical solution. Appropriate backup software packages make daily backups and write compatibility with all the predecessors. easier, while offering many tools that also allow management of the data to be stored, as well as data that has already been stored (archiving and HSM). The 7.3 Other Recording Methods necessary developments in the data media themselves also result in constant revisions and expansions of the various backup software packages. 7.3.1 QIC Drives Storage capacity requirements are growing faster than the capacities of individual media. Since purchasing more and more hard disks causes costs to skyrocket, With the serpentine format recording method, the 1/4“ cartridges are written there is an increasing need to utilise archiving, i.e. storage of data on more to at high speed in several cycles, each time on one track, parallel to the tape affordable but slower media. On the other hand this leads to high input in terms edge. This serpentine track format must also be traversed when reading data. of administration. 1/4“ tape cartridge drives work with the QIC (Quarter Inch Cartridge) format as a standard and record in streaming mode. If the transfer rate of a system 7.4.2 HSM Software exceeds the recording rate of the streamer, the tape will be written to continually (streaming), without pauses and adjustments, and thus be used to its full potential. HSM (Hierarchical Storage Management) software is able to perform this When the streamer‘s data buffer is empty it requests more data from the system. administration independently. Based on preset parameters like frequency of access, file size, file age etc., files are evaluated and if necessary transferred to ever more In order to fully utilise the tape on systems with slower transfer rates, the tape is affordable, slower media, for example, from hard disk to MO automatic changers, stopped and backed up to the end of the last recorded data. The next data block from there to tape libraries, and then on to tapes kept externally. In the latter case, is directly appended. Unlike helical scan technology, one advantage of the QIC the system administrator might be asked, if necessary, to insert the appropriate technology is that the tape always remains within its cartridge, resulting in much tape. longer life. A virtual file system always shows all the files to the user, not only those actually 7.3.2 Mainframe Drives contained on the hard disks, but also those that have been archived. When a file is opened, it is first copied back to the hard disk. The only thing the user notices 1/2“ magnetic tape drives are actually antiquated in today‘s world of high- is the correspondingly slower access time. In the worst-case scenario of the data performance data storage options. But they are still used since they are so widely being located on external tapes, waiting times when accessing a file might be as spread and are mainly used for exchanging data with mainframes. Since more long as half an hour or longer. Should the slower access time intimidate some compact and less expensive media are currently available, they will certainly users, then the fact that 80% of all data is accessed less then once a month might become less important in the future. 1/2“ tapes are usually written to with 9 serve as some consolation. Not even 5% of all data is accessed more often than tracks. once a week. On a large magnetic tape spool (with a tape length of 730 m), 23 MB can be stored using a write density of 800 bpi. This can be increased to almost 46 MB using a density of 1,600 bpi, to 92 MB using a density of 3,200 bpi, or 180 MB using a density of 6250 bpi. These are less than favourable ratings in comparison to storage media for magnetic tape cartridge systems that can record 10 GB and

340 transtec = ttec in the Netherlands and Belgium 8. OpticalStorage M has beenwritten. archiving. The stable, data mediaare butthedata very cannotbechangedonce it can becreated onCD small quantities catalogues, ofproduct orgraphic products software collections beam. I CDs) withoutgreat Data iswritten onto effort. theblankCD U 8.2 CD . terabyte range. make itpossiblefor asingleperipheral unitto managequantities ofdata upto the publishing ormultimediaapplications). Changersystems available onthemarket need to beavailable for CD direct well suited to applications where large amounts ofdata that are seldomaccessed changer timeperCDisaround 7to 10seconds. CD I can alsoread rewriteable CD drives offer anew feature: Furthermore, increase inaddition to inspeed, theobligatory present-day CD rotation speedisadjusted to prior reading. This yieldsaconstant data transfer rate. switches to thestandard in theinnerzone oftheCD The drives are distinguishedby avariabledata transfer rate andaconstant rpm technology ( of 7.8MB/s. Drives with read speedshigherthan14xare basedonpartial The present CD an audiodrive, whichis150 KB/s. A14xdrive, for example, canread at 2.05MB/s. The reading speedoftheCD CLV Technology information ispressed onto themedia. The ofCD production ofCD production drive) are usuallynotimportant. The media itself isextremelyinexpensive. M and therelatively slow access times(currently 85-120msec, dependingonthe quantities. The drawbacks oftheCD for since distribution, thereproduction software costs are muchlessfor larger 650 MB, theCD process.recorded ofthemanufacturing aspart With astorage ofapprox. capacity CD 8.1 CD also beingachieved. technologies,encompasses magneto-optical where improved storage is capacity created by DV New technologies, likeCD Optical diskshave found general acceptance intheform ofCD just like those distributed over-the-counter inlargerjust likethosedistributed over-the-counter quantities. With automatic CD n addition to the single-disk drives, CD n additionto thesingle-disk sing CDrecorders ( -R -R OMs are removable opticaldisksonlyfor reading data whichhasbeen OMs written withsuchaCDrecorder canberead CD onnormal n fact, itisbecoming lessexpensiven fact, to produce singleCD -R andCD -R OM Drives C onstant -R -R , are causingcontinuing growth inthismarket. This area also -R OM isanidealmediumfor ofdata, massdistribution e.g. OM drives work at upto 52xspeedandwithadata transfer rate s is usually only worthwhile inbatchesOMs isusuallyonlyworthwhile of100andgreater. -RW Drives -RW CD -R -R A -R), data canbewritten onto blankCD OM. Writeable CDsare alsosuitablefor long-term OMs ismuchlikethat oflong-playing records, the ngular CLV -R andCD multiread . As theoutermost track ofaCDisread, thedrive -R -RWs. ( C OM ismeasured inreference to thetransfer rate of V onstant elocity). elocity). -R OM access by acomputer (e.g. desktop -R -RW, aswell astheincrease incapacity . For CD OMs, i.e. to berewritten the inability -R L inear OM changersare alsooffered.typical The -R OM drives this signifies that they V elocity) methodwhereelocity) thedisc C edia AV Technology -R OM changersare especially -R OM usingalaser -R transtec = ttec in the Netherlands and Belgium and Netherlands the in ttec = transtec -R s (write-once OMs (write-once OM technology. -R OMs. Thus, -R OM drives, C AV -R ass

OM Furthermore, the DVdrives can read allcommon CDformats. For sometime by reducing theminimumlengthfor pitsand thetrack spacingofthewindings. isto bequadrupledinthefuture.capacity was extension achievedThis capacity 5.2 GB. usingbothsidesofthemediumandemploying By asecond data layer, this V from thevideoindustry, anewopticaltechnology emerged with For alongtime, ofCD thecapacity 8.3 DVrives drives having multiread (seetablefor capability compatibilities). to CD repeatedly, upto 3000times. These drives canalsowrite asingledata session CD rewriteable ( alone devices for duplicating CDs. CDs isat theforefront, CDduplicating systems have lately beenoffered asstand- back upinsmallormedium-sized networks. I changers that contain aCDrecorder there isacost-effective way to archive and R =read compatible W =write compatible to themagnetisation ofthematerial. Until now, to writing MOmediawas a very reflected laserbeam. The laserbeam informationreturns that directly corresponds The data mediacanbewritten to all withouttouching thediskorcausingany wear onit. soamagneticuse alaserto pattern heat theMOdisksurface, can be written to it, number ofadvantages over competing technologies. M M 8.4 M ofDVM Setup drives canalsoreadRW thecontents ofCD without acaddy, andcanalsoberead infuture DVdrives. Furthermore, theDV other hand, have developed aDV perside;itworks withacaddy.a 2.6GBcapacity Sony, P standard. Toshiba, andP Hitachi, The unrivalledsuccess ofDVwas, however, halted dueto thelackofacommon as yet onlyprovide astorage of2.6GB/side. capacity now, DV ersatile agneto-optical data storage,agneto-optical representing opticalstorage technology, hasa -R media.CD agneto- D -R drives (recorders) have beenavailable onthemarket. These, however, isk, originally: Digital originally: isk, Video to themediacapacity Disk)whichextended edia CD Optical Drives -RW mediacanberead withcommercially-RW available CD -RW) isatechnology that allows amediumto bewritten to -RW) anasonic developed theDV double-sided -R -RW drive,-RW whichoffers a3GBcapacity, works OMs was considered sufficient. With pressure -RWs, whileDV f, however, orcopying production of , anddata isread through the hilips, andHP agneto-optical diskdrives agneto-optical -RA -RA M drives cannot. M drive with DV , onthe ( D igital -R OM 341 -

compendium slow process. The reason was that any writing process needed to process each WORM disks (Write Once Read Multiple) are optical disks which are written to block three times: the first revolution would erase the block, the second would only once but can be read any number of times. With the introduction of CD-R, write the block, and the third revolution would verify it. Therefore, in comparison however, WORM technology has almost become obsolete. to a read request, a write request would take two additional disk revolutions. 8.5 Outlook By now, all commercially available magneto-optical drives are capable of LIMDOW (Laser Intensity Modulation Direct Overwrite), and can be switched to this mode. CD-ROM technology has won a place for itself in the fields of both audio and LIMDOW means, that it is now possible to directly overwrite old data with new computing. The development of CD-R (once writable) and CD-RW (rewriteable) data in a single pass, without the need for an erase cycle first. The claim of lower will ensure that these technologies will be successful in securing a market share. performance in comparison to conventional removable disk drives, for example, The situation is somewhat different for the DVD (Digital Versatile Disk) as this which reduced the magneto-optical technology‘s attractiveness until now, is no technology may replace CD-ROM in some areas, since each disc‘s capacity is many longer valid. times higher.

In summary, the advantages presented by these technologies are as follows: The In parallel with the developments in DVD, an MO video disk is being developed. data media are small, lightweight, and not very expensive. The MO drives can be This so-called MO-7 (ASMO) is expected to have a storage capacity of 6.1 GB. At used by a computer just like a normal disk drive. No special file system software is the same time, drive manufacturers like Maxoptix have introduced new products required. Areas where it makes sense to use magneto-optical disk drives are, for to the market that double the current capacity to 5.2 GB. The optical data storage example, those where the data medium has to be removed from the computer (for market is developing rapidly. The OSTA (Optical Storage Technology Association) example, high-security applications) and the higher performance of Winchester assists in gaining an overview of present and future technologies: www.osta.org. disks in removable cartridges is not required (or you do not want to pay their higher price). In addition, MO disks provide quick access to backups without copying data onto other media, such as disks, before it can be accessed. MO changers make it possible to archive very large amounts of data. However, these changers will become less important in the future, as CD-RW and CD-R jukeboxes take their place.

342 transtec = ttec in the Netherlands and Belgium 9. time it will eventually discharge, just like an accumulator.an like just discharge, eventually Towill of it loss time preventthe ideal. not is setup capacitor the that is disadvantage densities. large with and cost relativelylow a at manufactured be can it that is technology this chip.of the advantageto The connections same the via other the after one transmittedare which of address,each column a and addressrow a into matrix the accordingto subdivided is element data a levels:of address The two on electrically controlledare modules these costs,related and connections I columns. rowsand of matrix a zero.in logical arranged area forcapacitors The discharged is it whereasinstance, for one logical a storeto charged is capacitor A time. certain a for it retain and receiveenergy accumulators,can like which, capacitors, in stored is information that is memories dynamic of characteristic A memory. main as used generally is memory of drives. tape type example,forThis be,would this tocontrast a of example extreme An cell. data every on time any DRA DRA 9.1.2 D PD used asmassorprogramme innotebooks, memories routers, network printers, can for instance simplifydriver programming. Flash canfor memories example be place internally, sothat thisversion that isaccessed to afact similarly ahard disk, from outside. Ontheotherhand, for theATA flashcards address conversion takes modules have a address space where„linear“ any address accessed canbedirectly modules:Linearflashand offlashmemory main types ATA flash.Linearflash available.Regardless appearance, oftheir exterior there aretwo (MC 10 years. SIMM,PCC limited to about100,000.I those ofdynamicmemories. The numberofprogramming is and delete cycles addresses at atime. access Read times, currently about100ns, are aboutdouble read out, whereas write anddelete processes canonlyoperate onblocksof isthat offlashmemories individual A characteristic bytes canbeaddressed and Flash M aremanufacturer) stored incoded format (e.g. SPDsonDRA (operating voltage, access times, bankset-up, error etc., correction, down to the moduleproperties moduleswhere theparticular included invariousmemory available, eachwithdifferent properties. electrical For example, EEPR to ultra-violet light aswithitspredecessor EPR rather thanbybe changedordeleted othermeans(e.g., electrically by exposure EEPR EEPR are EEPR inthiscategory of memories contents, i.e., thestored data, even ifthepower supplyisswitched off. Examples What allnon-volatile have memories incommon isthat theyare ableto retain their 9.1.1 Non-Volatile M I 9.1 M computer. terms are typical explained Some inthischapter. increasingly module difficult tochoosetheappropriate foraspecific memory New technologies, faster chipsanddifferent hardware configurations makeit of acomputer‘s CPUadecisive factor, butalsothesystem‘s memory. working ofacomputer,When judging theperformance notonlyistheprocessing power made to this basic architecture and they are briefly described below: below: described briefly are they and architecture basic this to made been havedevelopmentsFurther intervals. regular atrefreshed be must it data, n order to keep the chips physically small, and to reduce the number of number the reduceto and small,physically chips the keep to order n n principle, there are static, andnon-volatile technologies. dynamic, memory As, anddigital cameras. ), and Solid State), andSolid Floppy DiscC is a dynamic memory with a free-to-choose access. free-to-choose a with memory dynamic a is M M M OM istheterm usedto allnon-volatile describe whosecontents memories can OM : emory emory Technologies ynamic M D emory , E²PR ain ynamic OM: emories emories M R andom E lectrically lectrically emories emories ard (PCMCIA), C emory A n general, data isretained for aguaranteed of period ccess E rasable M ard (SSFDC emory emory ompact Flashompact ( OM andFlash memories. P rogrammable

) are amongthevariousforms of OM). There are several architectures CF) I f it is not used for a long a for used not is it f R ead- ard, M I t can be accessed at accessed be can t O M DIMMs). transtec = ttec in the Netherlands and Belgium and Netherlands the in ttec = transtec nly iniature C M emory emory OMs are

ard I ts

has been structured in such a way to retain referenced information even if a new a if referencedevenretaininformation to way a such in structured been has processed. Withfurther receivedand be can circuits data the on signals electrical the that so interval, still”finite “kept a during be must memory the data, processto needed is time of period certain a Since referenced.is informationaddress new when retained!), are which cell memory the of contents the with confused be to (not deleted is circuits data the of signal techniques.addressing particular development. with modules memory memories, FPM to respect With EDO set. chip its and used system the by supported be must access of reduced.is bank, memory the of outputs the at available is data beforethe needed time cycle the that transferred,so be to need addresses column the only access subsequent access.Forevery first the for P address.row unchanged FastThethe re-transfer to necessary not Therefore,is it same the (on row“page”). same the in located generally are data consecutive since changes which address column the only is it addresses memory consecutive address.However,column the then fortransferred, and normally is address row the accessed,first is processor.bank the byprocessed memory a When arecontents programmes,memory computercontiguous within Normally method. addressing special a using applications particular forreduced are times the improvementof “standard” FastThe FPM graphics.large e.g.,with blocks, transferredin be tohave data of amounts large very when advantages processes.SDRA all synchronisingfor pulse clock a SDRA time, definite a at valid be must datarequired the addresses.Sinceconsecutive four all of contents the executedon is operation same the else,anything do to having without and, command the with together addresses.subsequent three the and X address out Read following: the be could command typical chip.A memory the on operation in set areprocesses defined command, the on depending transmission”SDRA the to addresses.column and row the DRA of types all for as Just space.address the in data consecutiveaccessing of means or FPM like Just eklectronics. disk hard the in incorporated of introduction the to itself.comparable chip is step This memory the tocontroller SDRA in presentdevelopment technology.this Thesupport explicitly which systemscomputer in used be only SDRA or FPM to contrast in But modes.access their of and architectures memory and FPM Like SDRA value. per pulse one of cycle clock a with contentscell threefollowing the an to similar cell the of contents the outputs module address,the an transmitting After mode. Burst additional an to due out SDRA asynchronous last the improvementof further A BEDO times. cycle the reduce further to module. helps This memory the in requiredaddress next the load and worddata referencedthe process simultaneously to possible accessed.is is it address Thus age hard discs, which have the controller tailored to their special needs special their to tailoredcontroller the have which discs,hard IDE : , a typical SDRA typical a M, : F M E technology is not downwardSDRA compatible,i.e.,not is technology M S M market. the on succeed to opportunity the had never has it M, : B ast xtended xtended ode makes use of this fact. Row and column addresses are only transferredonly are addresses column and Row fact. this of use makes ode urst P ynchronous P age age E xtended xtended EDO M I n a similar way, memory access is accelerated by the use of use the acceleratedby is access way, similar memory a n M D access methods, SDRA methods,access EDO ode available with some memory modules is the result of an of result the is modules memory some with available ode ata ode , SDRA , DRA O D

access is executed with the consecutive transmission of transmissionconsecutive the executedwith is access M utput D was introduced at the same time as the as time same the atintroduced was BEDO the As M. ynamic ata

technology is another development on existing on development another is technology M is also performed during this addressing and, addressing this during performed also is M O

utput is just the relocation of part of the memory memory the of part of relocation the just is M which is the is which BEDO the is modules memory EDO R DRA nlike with previous technologies,a previous with Unlike “command andom I n this case, the start address X is transmitted is X address start the case, this n

I memories, the electrical electrical the memories, FPM of case the n memory. delaytechnology-limitedM The

A O ccess

f course, in order to do so, this mode so,this do to order in course, f technology offers a powerful powerful a offers technology M memories, the output stage output the memories, EDO EDO M emory emory -RA memories provide speed provide memories M modules are provided with providedare modules M , however,attachesM, also it represent a further further representa EDO

memories can memories M

I t stands t

EDO 343 ,

compendium DDR SDRAM Double Data Rate SDRAM The DDR SDRAM, normally known as SDRAM II, is a faster version of the SDRAM. It has been further developed so that it can read or write data on the rising and falling edge of the system clock and therefore doubles the transfer rate of the memory. The maximum transfer rate of this technology is 1 GB/s.

RDRAM: Rambus DRAM Encoding memory module Rambus DRAM is based on one of the completely new technologies developed by Rambus Inc. The RDRAM is in the position to meet very high demands due to A DIMM has contact areas on both sides of the board but in this case the the overall reworking and redefinition of the inner structure e.g. conductor path opposite contacts are electrically isolated from each other. Thereby, more electrical lengths, pin capacities and voltage fluctuations. The memory is addressed on connections are available to a given sized module (= twice as many as with a the increasing and decreasing flanks of the system clock. A one-channel Rambus comparable SIMM), so that more data bits or a larger address space can also be memory with 1.6 GB/s offers three times the performance of a 64-bit 100MHz used. Furthermore, the number of connections for the operating voltage and SDRAM module. The main advantage of the Rambus technology is that 2 or 4 control signals increases, providing advantages with regard to the design of the Rambus channels can be used in parallel. Thus with 2 channels a bandwidth of 3.2 board and consequently the module‘s electrical properties. DIMM modules have GB/s can be achieved, and with 4 channels a bandwidth of 6.4 GB/s is possible. already become standard components in the PC, server and workstation sector. However RDRAM has never established itself in practice.

9.1.3 Static Memories

Static memories are used as cache memories due to their high speed. Unlike dynamic memories, they do not require that memory contents be refreshed. Static memory cells consist of a circuit composed of various modules. Depending on Layout memory module the data to be stored, the circuit adopts a specific state and then locks itself, so that its state can only change if induced from outside. Static memory loses its data content if the power supply breaks down. It hence belongs to the volatile 9.3 Terms relating to Memory memories. Due to the more complex structure of their , the memory densities which can be realised in static memories are lower than for dynamic Cache memories. Furthermore, for a comparable memory capacity, they are considerably A cache is a relatively small high-speed memory, usually SRAM, which buffers more expensive than dynamic memories. frequently required data between the CPU and main memory. The CPU can access data in the cache much quicker than the data in the main memory. A small cache 9.2 Modular Structural Forms memory can therefore considerably improve the computer’s performance without incurring a large amount of extra expense. Modern computer systems require memory sizes which, in data width (i.e. the There are different levels of cache, according to the position of the memory in the number of data bits used) and address space depth (i.e. the number of data words data stream. that can be accommodated in the memory), cannot be realised with a single memory chip. Therefore, modules are built from single memory chips, which can Level 0 Cache: be electronically addressed almost as though they were large chips. The following Decouples the data stream between the various processing units within the CPU. widely used structural forms can be distinguished: Its typical size is between 1 and 128 Bytes.

SIMM: Single Inline Memory Module Level 1 Cache: SIMMs have contact areas on both sides of the board which are, however, Decouples the data stream inside the CPU from the outside; its size varies from a interlinked. Therefore a 72-pin SIMM has 144 contact areas, but only 72 of them few Bytes (128) to several KBs. are electrically relevant. Level 2 Cache: DIMM: Dual Inline Memory Module Decouples the data stream between the CPU and the memory bus connected to the main memory. Its size ranges from 256 KBs to several MBs.

Level 3 Cache: Decouples the separate system bus from the memory bus with multi-processor systems.

Parity/Non-Parity External influences (electromagnetic fields, voltage fluctuations, radioactive radiation, cosmic radiation, etc.) can cause individual data bits to „invert“: Thus a logic zero becomes a one and vice-versa. The stored data is therefore altered. Non-parity modules only store data and do not offer any means of error detection. Parity modules, on the other hand, store data and checksum information. This is derived as follows: If the number of set bits is even, the parity bit is likewise set, whereas if the number is odd, the parity bit is cleared. This parity technique helps detect one-bit errors. Since no error correction occurs, when a parity error occurs most systems do not report an error message. Normally, 1 parity bit is stored per data byte (8 bits).

344 transtec = ttec in the Netherlands and Belgium data bus. data set. chip the through (memory) memory between bus system the is bus memory The M place.procedure take this notice not will user audio of correction error for Ascorrected. and detected be can errors bit individual checksums, these of help sections. the With memory separate the to (similar blocks in formed are checksums as referredto „Error also occasionally is fly.method the on underlying correctedThe be can errors bit The ECC emory Bus emory : process is able to detect both one and two-bit errors. two-bit and one both detect to able is ECCprocess E rror

C hecking and hecking D etection and etection C orrecting orrecting C orrection“ (EDC orrection“ I t consists of the address bus and the and bus address the of consists t and I/O

CR method) and are stored in storedare and method) C ). that addresses the main the addresses that CPU U sing particular algorithms, particular sing transtec = ttec in the Netherlands and Belgium and Netherlands the in ttec = transtec M s, the software software the CDs, oreover,one- manual should be consulted.be should manual differentmanufacturers.from and differentcapacities of modules memory with fitted be can differentspecifications,banks system on However, bank. depending memory given a within identical be must modules The with fitted be should system. servers Whereverpossible, particular their of specifications the accordingto used be to memory manufacturer.the defines manufacturer The I for whatWhat computer? memory n principle, the type of memory which can be used is specified by the hardwarethe by specified is used be can which memory of type the principle, n ECC.

I f in doubt, the system the doubt, in f 345

compendium 10. Communication

10.1 Definition 10.3.2 Two-way Communication

Communication takes place between senders and receivers. Communication can As opposed to one-way information, in two-way communication information is be one-way or two-way and serves for the exchange of information. When such exchanged concurrently. Those taking part in the communication are at one and information takes the form of data, text, images or voice, one talks of telecommu- the same time: sender and receiver. nication. 10.3.3 Interactive Communication 10.2 Prerequisites for Successful Communication Interactive communication is a developed form of two-way communication. The Three essential prerequisites must be met for successful communication: concurrent exchange of information interacts with information relating to the last - There is at least one sender and at least one receiver exchange. A dialogue develops between the parties to the communication. - Sender and receiver are linked to a joint medium - Sender and receiver speak the same language. 10.3.4 Open Communication

10.3 Communication Types One talks of open communication (open system interconnection) when parties to the communication follow standardised communication rules. By standardising the 10.3.1 One-way Communication communication rules, all potential parties to the communication can ensure that they can communicate with as many other potential parties as possible. Hence the In the case of one-way communication, information is exchanged in one direction possibility of communicating is open to all potential parties. only. The sender transfers its information to the receiver through the joint medium. If, as a consequence of having received information, the receiver answers the sender, this in turn is regarded as separate, one-way communication. The receiver becomes the sender and vice versa.

346 transtec = ttec in the Netherlands and Belgium 11. Standards

11.1 Definition of Standard generating, transporting and distributing electrical power, domestic electrical installations, telecommunications and radio engineering and information A standard is understood to be a universally valid definition in answer to a set technology. of facts. Before having been adopted, a standard has always passed through an approved standardisation procedure recognised by law, which is made up of 11.3.3 International Telecommunication Union (ITU) several instances and process steps. The prime criteria for a standard are technical maturity and user benefit. The International Telecommunication Union (ITU) was founded in Paris by 20 states on 17.5.1865 and since 15.10.1947 has been a suborganisation of the 11.2 Types of Standards United Nations headquartered in Geneva. The ITU is a globally active organisation in which governments and the private sector of telecommunication coordinate the It is important to differentiate betweenstandards, industrial standards, operation of telecom networks and services. manufacturer-specific standardsand recommendations. 11.3.4 European Telecommunications Standards Institute (ETSI)

Standards have passed through the defined process of standardisation (see definition of standard). The telecom standards valid for Europe are issued by the European An industrial standard is a non-standardised procedure or operation in answer to Telecommunications Standards Institute (ETSI) founded in 1988. The institute a set of facts, which is manufacturer-independent and has as a rule proved to be has its headquarters in Sophia Antipolis, a research centre in the south of correct by having been successfully implemented in practice for many years. France. The ETSI has approx. 800 members from over 50 countries. The ETSI The manufacturer-specific standard differs from the industrial standard in that here European standards of telecommunication are mostly based on the corresponding manufacturer-specific procedures or operations in answer to a set of facts have international recommendations (e.g. of the ITU). The ETSI modifies these standards been proven and tested in practice. in line with the European requirements. The recommendation is the most non-committal type of standardisation. Neither manufacturer nor user is obliged to conform with it. 11.3.5 Institute of Electrical and Electronical Engineers (IEEE)

11.3 Important Organisations for Standardisation in IT The Institute of Electrical and Electronical Engineers (IEEE) in New Jersey, USA, addresses itself to standardisation tasks, among others, and is an association of 11.3.1 International Standardisation Organisation (ISO) chiefly American engineers. The IEEE is divided into various working groups, which are formed temporarily until a recommendation for a standard has been drawn The national standardisation institutes from 118 countries work together in the up. The working group 802, for example, attends to the standardisation of local International Standardisation Organisation (ISO) based in Geneva. The objective of networks and has made major contributions including such for defining access ISO is to facilitate the international exchange of goods and services by standards procedures and security protocols. of worldwide uniformity and to promote scientific, technical and economic co- operation beyond national frontiers. The ISO is responsible for the best-known 11.3.6 ANSI standard from the field of data communication and is still developing it: TheO SI reference model. IEEE is a national standardisation body in the USA. ANSI (American National Standards Institute) develops and publishes standards. ANSI is non-profit-oriented 11.3.2 International Electronical Commission (IEC) and works independently of the government. It is supported by over 1000 trade organisations, professional associations and companies (comparable with DIN in The International Electronical Commission (IEC) was founded in 1906 and Germany). later affiliated with the ISO. It has more than 600 working groups dealing with ANSI is the US representative and voting member of the ISO. The members, standardisation. The standards drawn up by the IEC enjoy worldwide validity. The manufacturers, research groups, other standardisation bodies and other interested IEC is based in Geneva, the national standardisation institutes operate in over a parties paying dues are organised similarly as in ISO. hundred countries. The IEC is a recognised standardisation organisation in the field of electrical engineering and electronics. Its field of reference includes electrical safety, electric strength and insulation, electrical components, installations for dium n compe

transtec = ttec in the Netherlands and Belgium 347 12. The OSI Reference Model

12.1 Introduction communication equipment (DCE). The bit stream is conveyed via a communication channel. Examples of characteristics will be set forth below. Mechanical: these

The OSI model serves as a reference model for all open and standardised systems. may be the dimensions of the connectors or the arrangement of the pin OSI stands for Open System Interconnection and provides an abstract model as a structure. Electrical: the voltage level on the circuits, e.g. according to the X and basis for interpreting existing systems and as a reference for the definition of new V recommendations of the ITU. Functional: the assignment of the functions of protocols. To be able to understand the development of the OSI model, you have individual pins. Procedural: the general rules for the use of interface circuits. to bear in mind that the structure of the IT world before 1980 was non-uniform, governed by manufacturer-specific standards which were incompatible with one Layer 2, Data Link Layer another. Any modifications in the interest of achieving compatibility were costly The Data Link Layer has to ensure transparent data exchange. The bits streams are and involved a lot of effort, a situation corrected by the OSI reference model, combined to form so-called frames and forwarded to the Network Layer, passing which is based on an ISO standard and was adopted by the ITU in 1984. The through the processes of error detection and error correction. objective of the OSI reference models is the unification of standards to enable communication between open and standardised systems. The OSI reference model Layer 3, Network Layer is subject to ongoing, flexible updating, i.e. old elements are extracted and new The Network Layer realises the data transport between end systems, via ones added to prevent the reference model from becoming obsolete. any intermediate nodes, by means of routing. The data units are packets in connection-oriented transmision and datagrams in connectionless transmission. 12.2 Structure Therefore entire networks can be interconnected, the basic requirement being a unique address. X.25, Frame Relay and IP are typical protocols of the Network

The OSI reference model is a layer model, this means the overall system is divided Layer. into subsystems: layers. The individual layers conform with a specific structure. Subsystems of the same hierarchy form a layer, the layers are superposed, Layer 4, Transport Layer whereby a lower-level layer serves the communication function of the higher-level The Transport Layer realises the data transfer irrespective of the underlying layer. Further, each layer provides certain services both within that layer and for network and provides reliable data traffic. Address-mapping, flow control lower-level layers. between hosts with different speeds and multiplexing of transport connections in networks are typical functions of this layer. The best known protocols of this layer In the OSI reference model there are seven layers in all, which can be divided are UDP (User Datagramm Protocol) and TCP (Transmission Control Protocol). into two superordinate categories: the transport-oriented layers 1 - 4, which are Layer 4 simultaneously forms the transition to the application-oriented layers. responsible for transmitting bits, and the application-oriented layers 5 - 7, which are responsible for the control of the operation and the nature of information Layer 5, Session Layer presentation. The Session Layer realises the session, i.e. the communication of processes. Further, it organises the data exchange with added processes such as dialogue The layers communicate with one another according to specific rules, the and communication control as well as synchronisation in case of systems crashing. protocols, the functions of elementary communication being called service primitives. The service primitives are classified into four families per layer: Request, Layer 6, Presentation Layer Indication, Response and Confirmation. In a communication between A (service The Presentation Layer establishes the communication to application processes user) and B (service provider) interconnected via layer N for service provision, A of heterogeneous computer systems. It deals with syntax only, the uniform issues a Request for a service from B and the Indication informs B of this. Then processing and presentation of the data. This entails, for example, the comes the Response from B, which represents the acknowledgement of the transformation of different data formats into a standard form, the so-called code preceding Indication, and finally the Confirm as the provider‘s acknowledgement conversion. of the Request. Layer 7, Application Layer 12.3 Layers The Application Layer provides services for the external use of computer networks and is the most comprehensive layer. Functions may be: identifying

Layer 1, Physical Layer communication partners, granting access rights and specific service qualities. The Physical Layer defines the mechanical, electrical, functional and procedural characteristics of physical links between data terminal equipment (DTE) and data

348 transtec = ttec in the Netherlands and Belgium 13. Transmisson and receiving devices.receiving and sending of control time the for method a is and transmission data asynchronous and synchronous layer,in link used data is the to applies Synchronism 13.3 Synchronism assumed. is states two of one only that signifies binary connection, dual-system. called used.also areprocedure is This characters two only signal. value-discrete and time-discrete a is I play. into come modulation phase and modulation amplitude modulation, frequency of processes modulation the voltages.Here the in changes constantrepresented by signal. value-continuous and time-continuous a is it i.e.value, maximum positive and negative the I value-discrete of speaks one permitted,are values of range the in values signal certain only if of speaks one permitted,are values of range the in values of speaks one times all at present not is value of speaks one times, all at present is value signal I 13.2 Signal entirety.its in communication data the describe thereby and another one complement they rathermore network, a in individually regarded be cannot categories these that is point below.important with An dealt be will transmission data of categories duplex. various full These and duplex half-duplex, transmission character synchronism the of characteristics. character The various accordingto categorised be can networks in transmission Data 13.1 I n the case of a of case the n an of case the n classes. various into subdivided be can signals time-dependent principle, n

ntroduction , i.e. asynchronous/asymmetrical or synchronous/symmetrical, the synchronous/symmetrical, or asynchronous/asymmetrical i.e. , .

digital signal digital analogue signal analogue

, i.e. bit-serial or bit-parallel and the and bit-parallel or bit-serial i.e. , I n analogue data transmission, the information is then is information the transmission, data analogue n , the signal can assume only certain values, i.e. it i.e.values, certain only assume can signal the , , the signal can assume any value between value any assume can signal the , M signal ethods and Techniques , i.e. analogue or digital, the digital, or analogue i.e. , I n the depiction of digital information digital of depiction the n time-discrete time-continuous value-continuous mode . I f all the signal the all f transtec = ttec in the Netherlands and Belgium and Netherlands the in ttec = transtec simplex, , if a signal a if , I n this n , I f a f the direction. one in only place take can communicationreceiver, the or that so sender either be only can However, channel. communication commonparty each duplex. duplex/full and half-duplex simplex, being these operation, of differentmodes three between I 13.5 Operation parallel. in time same the at transmittedare characters sequence. time in I differentiatestransmissionbetween character The 13.4 Character Transmission transmission. next the forprepare to assumed is state idle an stop-bit the After inbetween. slot time fixed a with stop-bit a and start-bit a by defined is frame the means I header.frame special a by marked is frame the of frame. a beginning intoThe combined is bits data of number free A edges. data the via out carried is generator clock the of synchronisation the kept; receiverare and sender between intervals time exact wherebytransmission, character-oriented I place in two directions simultaneously. directions two in place take can communication the that so time, same the receiverat and/or sender be channels.communication common two through connected are possible. is time a at direction one only but directions, both in place take can communication the that so receiver,is who and sender is who agreeto have parties the communication to prior time.that signifiesa Thisat roles these of one perform only can but receiveror sender be can party each howeverherechannel, communication n sender-receiver communication in networked networked in sender-receivercommunication n transmittedare characters the of bits the transmission, bit-serial of case the n n n asynchronous data transmission data asynchronous transmission data synchronous half-duplex mode, two parties are likewise connected through a common a through connected likewise are parties two mode, n the other hand, in bit-parallel transmission the bits of the of bits the transmission bit-parallel in hand, other the On I n the n simplex I n the n mode two parties are connected through a through connected are parties two mode distinction is made between bit-oriented and bit-oriented between made is distinction the start-stop principle is employed.is principle That start-stop the duplex or full-duplex mode, both parties parties both mode, full-duplex or T systems, distinction is drawn is systems,distinction IT

bit-serial and ach party can party Each bit-parallel . I n 349

compendium 14. Personal Area Networks-PANs

14.1 Definition provides object exchange services similar to WWW Hyper Text Transfer Protocols and IrDA Lite, which provides methods of reducing the size of IrDA code. There is

The term (PAN) covers the communication of devices of a also the IrTran-P, which provides an image exchange service used in digital image single or small number of users within a range of about 10 metres. As a rule three capture devices/cameras. The IrMC protocol plays an important part in mobile areas can be defined. The linking of peripherals, the linking of external operator communication, providing the specifications for data exchange between such devices and the linking of multiple computer systems for data transmission. devices. Finally the IrLAN protocol describes the interface between IR networks The latter can be disregarded as it borders on the classical LAN. Since most and an LAN (). PANs function wirelessly, one also speaks of Wireless Personal Area Network (WPAN), which is virtually a synonym. The two technologies most often used in 14.2.1.2 IrDA Control this connection are IrDA and Bluetooth, which will be discussed in further detail below. IrDA Control is a communication standard enabling wireless periphials to communicate with many different intelligent host devices. 14.2 Transmission Methods Protocols

14.2.1 IrDA (Infrared Data Association) The necessary protocols of an IrDA Control implementation include: PHY (Physical Signalling Layer), MAC (Media Access Control) and LLC (Logical Link Control).

The term IrDA (Infrared Data Association) covers a network concept for short range, line of sight, point-to-point cordless data transfer, also referred to as IrDA Physical Signalling Layer (PHY) Data. IrDA Control, on the other hand, provides the possibility of linking multiple The range of operation of IrDA Control is comparable with that of present-day peripherals for point-to-point or point-to-multipoint cordless data transfer without uni-directional infrared systems and is at least 5 m. In the same way as with IrDA direct line of sight. It must be noted that IrDA Control provides lower speeds than Data, bi-directional communication is the basis for all standards, a data transfer IrDA Data. rate of max. 75 KB/s being reached. The data packets are checked using a CRC (Cyclic Redundancy Check) procedure (CRC-8 and CRC-16).

14.2.1.1 IrDA Data Media Access Control (MAC)

Since IrDA Data was passed as a standard in 1994, well in excess of 300 million The Media Access Control enables a host to communicate with mutiple peripheral electronic devices with these interfaces have been in use. This is due simply to devices, but with a maximum number of eight periphals simultaneously. the fact that there is an increased demand for communication between electronic Furthermore, a fast response time and low latency are ensured. devices, ranging from established devices of daily use to newer devices emerging from the development of the IT sector. Logical Link Control (LLC) The Logical Link Control ensures reliability features providing data sequencing and Protocols re-transmission when errors are detected. The mandatory protocols of an IrDA Data implementation include: PHY (Physical Signalling Layer), IrLAP (Link Access Protocol) and IrLMP (Link Management 14.2.2 Bluetooth Protocol). Bluetooth is a solution for the simple connection of peripherals via a wireless Physical Signalling Layer (PHY) interface. The standard was introduced by the Ericsson company, which gave The range for continuous operation is 1 metre to 2 metres. In a low power Bluetooth its name. The Danish King Harald II, who reigned around 1000 years version, ago and is still held in high regard in Denmark, was called Bluetooth. The the range is 20 cm to max. 30 cm for communication between low power and standard is promoted by the BSIG (Bluetooth Special Interest Group) and was standard power versions. Bi-directional communication is the standard of all introduced for the first time in 1998. Along with Ericsson, nine companies form specifications, it being possible to attain data rates of 9600 B/s Infrared (IR), 115 the so-called Promoter Group, while in the meantime a significantly larger number KB/s Serial Infrared (SIR), 4 MB/s Fast Infrared (FIR) and 16 MB/s Very Fast Infrared of companies belong to the lobby. (VFIR). The data packets are checked with a CRC (Cyclic Redundancy Check) procedure (CRC-16 and CRC-32). 14.2.2.1 Bluetooth - Basics

Link Access Protocol (IrLAP) Bluetooth operates in the 2.4 GHz ISM unlicensed band, the same frequency The Link Access Protocol provides a device-to-device connection for reliable, band as used by the wireless LAN Standards IEEE802.11b and IEEE802.11g. There ordered data exchange. It provides functions for establishing and clearing are three different power classes: Class 3 operates with a sensitivity level of -70 connections and for fault and flow control. dBm, an ouput power of 1 mW, a theoretical range of 10 metres and a network size in the piconet range. This is the class in which most Bluetooth devices today Link Management Protocol (IrLMP) operate. Even if according to definition this is the range of PAN, Bluetooth is The Link Management Protocol provides multiple channels via an IrLAP however not limited to this level. The output power can be significantly increased connection. It also provides functions for multiplexing for the IrLAP connection. by added power amplifiers. Class 2 operates with a sensitivity level of 4 dBm, an The Information Access Service (IAS), provides protocol and service discovery, output power of 2.5 mW, a theoretical range of 30 metres and a network size forming the basis for the communication of IrDA components. in the nanonet range. However, maximum power is achieved in class 1 with a sensitivity of 20 dBm, an output power of 100 mW, a theoretical range of 100 Optional Protocols metres and a network size in the micronet range. Networks of classes 3 and 2 are Further optional protocols are on layers 4 and 5 of the IrDA Data protocol to be associated with the PAN range, while class 1 networks form a borderline stack, these being, for example, Tiny TP which provides flow control on IrLMP case to the LAN. It must be noted, however, that in the present-day Bluetooth connections with an optional segmentation service. IrCOMM provides COM (serial solutions, data rates of 1 MB/s max. can be realised. Future expansions are already and parallel) port emulation for legacy COM applications. Further, OBEX, which being worked on with the aim of doubling the data rate to 2 MB/s.

350 transtec = ttec in the Netherlands and Belgium the frequency hopping. That means that a master and seven slaves at most can most atslaves seven and master a hopping.that means frequency That the of control the forresponsible being and master as acting network the in device active number,first identification the long 48-bit a using themselves identify another.one with devicescommunicate Bluetooth can The devices Bluetooth eight than more no piconet a In piconet. be to said is size network the 3 class In 14.2.2.2 Bluetooth -Network out. carried is transmission data during encryption bit 128- to 8-bit an and authentication 128-bit a established, is connection the When methods. encryption employing 3 mode only modes, security three has Bluetooth realised. be can direction upload the in KB/s 57.6 and direction download the in KB/s 723.2 of rate transmission data a transmission, asynchronous of case the In KB/s. 433.9 of rate transmission data a providestransmission Synchronous selected. be can transmission asynchronous and synchronous principle In each. KB/s 64 with channels voicethree to up operateto possible being it transmission, data and voicefor channel wide-band a providesBluetooth each. MHz 1 of steps 79 in second a times 1600 switched of method fast GHz. The 2.480 to GHz 2.402 from being range frequency operating exact the spectrum, ISM GHz 2.4 the in operatesBluetooth mentioned,already As F requency requency H opping S pread S pectrum (FHSS) is used for access. Frequency is access.Frequencyfor used is (FHSS) pectrum

transtec = ttec in the Netherlands and Belgium and Netherlands the in ttec = transtec market penetration and a more sophisticated technology.sophisticated more a and penetration market wider significantlyhave IEEE802.11 accordingto standards industrial the addition from case WPANborderline representsthe this regarded,as critically TO In WLAN. be to has also transmission data networked for Bluetooth of band. use ISM The the in likewise standards IEEE802.11g or IEEE802.11b the accordingto operating componentswith interferenceWLAN be may there that means band frequency GHz 2.4 the However,fulfilled.of be use can the different manufacturers from devices Bluetooth with interoperability an that ensures Bluetooth name brand The use.widespread to obstacle no be will there met, are components Bluetooth for solution chip single economical, very a of targetsdevelopment the If sub-areas. in greatly varies it however,as market, qualificationrequires the this on acceptance rapid the is standard open an such peripherals.of advantageof The connection PANsimple the permits in and asset segment an is technology Bluetooth The 14.2.2.3 Bluetooth -Advantages/Drawbacks piconets. two between intermediary as acts slavecommon a doing, so In scatternet. so-called a form to piconets multiple combine to possible is Furthermore,it necessary. if master the through activated be then can and slavespassive as parked be also can devices However,Bluetooth piconet. 255 a to in up operate

351

compendium 15. Local Area Networks - LANs

15.1 Definition the next station following this one, the defective or inactive station is skipped and no longer included. Networks based on the token principle are very useful for In the age of information technology, networks exist that span the globe. The applications that require guaranteed response times. best-known network is the Internet, which is becoming more and more popular. It is important that all of the advantages of a network apply to both small and large 15.3 Ethernet and 802.3 LAN networks. Networks that have a scope that is limited to a company‘s premises or a building are called a Local Area Network (LAN). A LAN contains at least two It is important to differentiate between the synonyms Ethernet and 802.3 LAN. networked PCs and can stretch to cover a multitude of PCs and servers. The goals Ethernet was originally developed by DEC-Intel-Xerox in 1980 with the CSMA/CD of this network can be divided into four important sections. These are “data access mode. This standard was taken over by the IEEE (Institute of Electrical and interlocking”, i.e. access to remote data, and “function interlocking”, i.e. access Electronics Engineers) and then released as the ISO standard 802.3 LAN in 1990. to computers that provide special services, such as servers. The last two are “load The only essential difference between these standards are their frame formats, interlocking”, i.e. the load is spread out over several computers, and “availability but they can co-exist in one network. The present-day network components are, interlocking”, which guarantees failsafe operation with redundant computer in principle, manufactured only according the 802.3 standard, still referred to systems. When talking about a LAN, you must differentiate between wired as Ethernet. The logical topology of an Ethernet is a bus system, whereby the and wireless structures. If it is a wireless structure, then it is also called a WLAN physical topology depends on the various network standards and can also take on (Wireless Local Area Network). The following will first explain the classic LAN. other forms, such as a star structure.

15.2 Accessing Layered models In the case of an 802.3 LAN layer 2, data link layer, is divided once more into a

Since several computers access the same transmission medium, concurrent MAC sublayer and a LLC sublayer. The MAC sublayer (Medium Access Control) and reciprocal access must be regulated. Networked IT systems can be easily controls access to the transmission medium and the allocation of the transmission classified according to such access modes. Essentially one can distinguish between capacity, whereas the LLC sublayer (Logical Link Control) sets up the connection two access methods: the non-deterministic, random access mode and the to protocols in higher OSI layers. There are three types of LLC sublayers. Type 1 deterministic, sequential access mode. CSMA/CD (Carrier Sense Multiple Access provides a connectionless transmission, type 2 a connection-oriented transmission / Collision Detection) is the best-known method of a non-deterministic access and type 3 a mixed form with connectionless and connection-oriented mode, which is used in Ethernet and 802.3 LAN. Deterministic access modes transmission. generally work according to the token principle, e.g. token passing (token ring I), early token (token ring II), time token (FDDI) and polling mode (VG Any LAN). Packet format CSMA/CD and token passing will be explained in more detail below. Ethernet is a network that conveys packets, i.e. the data that is to be transmitted is divided into small units, packets or even frames. Every packet searches for its

15.2.1 CSMA/CD own path through the network. They are only put back together at the end of transmission. The maximum frame length for an 802.3 LAN packet is 1518 bytes, CSMA/CD stands for Carrier Sense Multiple Access / Collision Detection. Carrier the minimum frame length is 64 bytes. Larger packets are usually required in Sense (CS) stands for the monitoring of the transmission medium and listening modern-day networks. These are called jumbo frames, which describe packets for communication. If a station would like to send something, it waits until the that are larger than 1500 bytes. Due to downward compatibility, an increase medium is free and then for a further time span, before it starts transmission. in packet size has been avoided in current Ethernet standards, which causes Multiple Access (MA) stands for equal, competitive access to the shared the performance of high-speed networks to suffer. Jumbo frames can be up to transmission medium. Collision Detection (CD) detects collisions by monitoring 9000 bytes, beyond that there are restrictions against larger packets, due to the and listening to the transmission medium. If a station realises that a collision 32-bit CRC process that is used in Ethernet. CRC is an abbreviation for Cyclic has occurred while transmitting its packages, it sends off a so-called jamming Redundancy Check, which performs a check using generator polynoms that is signal and all of the sending processes are aborted. The station will only try to output in the frame check sequence (FCS) field. Another increase would be useful, transmit again once a specific time span and a randomly set time span have both but it not expected in the foreseeable future. Components that support jumbo passed. Due to this, Ethernet is called a collision-oriented network, since collisions frames are usually identified specifically, however, many components can support occur depending on the design of the system and the load on it. However, it is jumbo frames to a certain degree. important to realise that CSMA/CD can only function if the doubled maximum round trip delay (RTD) between the stations that are furthest apart is shorter than 15.4 MAC Addresses the transmission time for the smallest permissable package in a 512 bit Ethernet. The 5-4-3 repeater rule has arisen in this context and is still important today. MAC or LAN addresses are the elementary addresses that are used to clearly It states that a maximum of five segments with four repeaters and three inter- identify components in a local network. There are generally two different kinds repeater links (IRL) can be coupled in an Ethernet. of addresses, a 16-bit and a 48-bit variant. In the 16-bit address, the first bit identifies whether it is an individual or group address, the other 15 bits are used

15.2.2 Token Passing for further identification. The group address can be a multicast group address or a broadcast address. However, 48-bit addresses are used almost exclusively

In contrast to CSMA/CD, token passing is a set mode where collisions do not today. The first bit in these addresses is also used for identification of individual or occur at all. Every station can only send if it has a specific bit pattern, the token. group addresses. In addition, there is a second bit, which identifies whether it is This token gives it authorisation to send and is passed from station to station a global or local address and the other 46 bits are used for further identification. to a physical or logical successor. The maximum sending time is limited, which With global addresses, the manufacturers of controller chips have to buy the first means that response time is calculable in such networks Here a station checks 3 bytes of the MAC address from the IEEE. The manufacturers generally have a after the token has been passed if its successor has received the token correctly. defined spectrum of addresses that they can use for their controller chips. The If it receives the token without any identification of its successor, then the station remaining 3 bytes of the MAC address are assigned by the manufacturer. This sends the token back on its way. If a second attempt also fails, then the system guarantees that the addresses are unique worldwide and that there are not any assumes that this station is defective or not active. After the token has checked double addresses, which can lead to complications in a network. Since no one can

352 transtec = ttec in the Netherlands and Belgium 15.5.5 100BASE-TX installations. new in included not is it but while, short a just for extensively used been has concentrators.standard and BASE-FL stations 10 between The as well Optic (FibreFOIRL to expansion an is (15.9). standardchapter The later a in discussed be will standards cable connectors. various ST The use often which medium, transmission the as used are cables duplex Fibre-optic cabling. fibre-optic-based with m 2,000 of length segment maximum a with transmission baseband Mbit/s accordingto BASE-FL 10 standard The 15.5.4 10BASE-FL installations. new in included not is it but while, short a just for extensively used been has amplifier.the BASE-Tas 10 used standardThe also concentratoris the shape, star a into stations concentratethe to used is component active topology,one star i.e. chapter.later a BASE-Tthe in 10 uses physicallydiscussed The be will standards cable connectors. various RJ-45 The with along medium transmission the as cabling.copper-based with m 100 of length segment maximum a with transmission baseband Mbit/s BASE-T10 accordingtostandard The 15.5.3 10BASE-T installations. new in included not is it but while, short a just for extensively used been connector.has standardthe BASE-2 as 10 used The is adaptorBNC T the and segment per connected be may stations 30 of maxumim A stations. two between switched be mayrepeaters 4 of maximum a and m 0.5 is stations two between length segment minimum topology,the whereby medium. transmissionThin a as cable coaxial ohm as Thin known also is BASE-2 10 m. 185 of length segment maximum a with transmission baseband Mbit/s accordingto BASE-2 10 standard The 15.5.2 10BASE-2 installations. new in included not is but past, the in extensively used A the of length maximum (A interface unit attachment the wherebysegment, per connected be may stations 100 of maximum A (IRL). inter-repeaterlinks 3 of together.maximum connected area thatrepeaters in results 4 overThis m 500 of Thick a consideration.that into means taken This be to has rule repeater 5-4-3 the medium. transmissionThick a as cable as Thick known also is BASE-5 10 m. 500 of section transmission maximum a with transmission baseband Mbit/s accordingto BASE-5 10 standard The 15.5.1 10BASE-5 below.detail more in described are practice into put be possibly will and practice in already are that standards important m. most The500 of length segment an refersto accordingto BASE-5 10 standard the of nomenclaturetypes. the Thus connector as well as length, segment maximum mode, transmission speed, of from implementations of variety a is There 15.5 E locally. the from addresses the buy forcedto be The standard 100 BASE-TXaccordingto 100 standard The et may have a maximum of 5 segments with a maximum segment length segment maximum a with segments 5 of maximum a havemay Net thernet Standardsthernet I nter Repeater Link) and defines the connection between concentrators,between as connection the defines and Link) Repeater nter

E thernet with 10 with thernet C opper twisted pair cables of various standards are used are standards various of cables pair twisted opper -5 standard has been has standard BASE-5 m. 10 The50 is cable UI bit/s baseband transmission with a maximum a with transmission baseband Mbit/s et or Net et, since it uses a thick RG8 50-ohm coaxial 50-ohm RG8 thick a uses it since Net, Cheaper 802.3i refers to an refersto IEEE802.3i 802.3a refers to an refersto IEEE802.3a an refersto IEEE802.3 802.23j refers to an refersto IEEE802.23j et physically uses the bus topology and topology bus the uses physically Net 802.3u refers to an refersto IEEE802.3u IEEE E thernet, which mainly differ in terms in differ mainly which thernet, , the adresses can also be managed be also can adresses the , et, since it uses a thin RG58 50- RG58 thin a uses it since Net, et physically uses the bus the uses physically Net ) connector is used. is The connector UI) E E E thernet with 10 with thernet thernet with 10 with thernet transtec = ttec in the the in ttec = transtec thernet with 10 with thernet E E thernet with 10 with thernet thernet with thernet IEEE802.3, N etherlands and Belgium and etherlands T tnad is standard BASE-T 1000 The amplifier. the as used also is concentrator the shape, star a into stations the concentrate to used is component active one i.e. topology, star the uses physically BASE-TX 1000 The (15.9). chapter later a in discussed be will standards cable various The connectors. RJ-45 with along medium transmission the section. terminal the in 1000 with to according BASE-T 1000 standard The 15.5.7 1000BASE-T LWLin installationsenvironments. new in also used, widely is standardprocess. BASE-FX token 100 time The the of basis the on works which (FDDI), interface data distributed fibre the with used is chapter.laterBASE-FX 100 SC, ST,use often which medium, transmission the as used are cables duplex optic concentrators.between Fibre- m 2,000 concentratorsand and stations between 100 accordingto BASE-FX 100 standard The 15.5.6 100BASE-FX installations. new in often very used, widely BASE-TXis amplifier.the as 100 used The also concentratoris the shape, star a into stations concentratethe to used is component active topology,one star i.e. the uses BASE-TXphysically(15.9). 100 chapterThe later a in discussed be will standards cable connectors. various RJ-45 The with along medium transmission the as used cabling.copper-based with 100 Semiconductor, which is also often referred to in descriptions. protocolin Thereferred to often also Semiconductor,is which 10 Fast The 15.5.10 Auto-Negotiation area. this in acceptance gained has and sector LWL the primary in cabling I of use the without and connction other.point-to-point each a to compatibleuses standardonly Thisare that solutions manufacturer-specific usually are these but bridged, be can distances fibres.Larger mode single micron 9/125 for maximum m 5000 and fibres multimode micron 50/125 or 62.5/125 for maximum m 550 of distance a bridge can cable, fibre-optic the on depending and, used is nm 1300 wavelengthof a 1000 with accordingto standard BASE-LX 1000 The 15.5.9 1000BASE-LX LWLnew installations.in used and accepted widely is standard this match, copper the Besides used. be also mayand VF-45 RJ with connection to-point point- a fibres.uses multimode standardmicron This 50/125 for maximum m 550 and fibres multimode micron 62.5/125 for maximum m 275 of distance a bridge can cable, fibre-optic the on depending and, used is nm 850 wavelengthof a 1000 with accordingto standard BASE-SX 1000 The 15.5.8 1000BASE-SX installations. new in used is and accepted widely more and more becoming is standard This wires. of pair ever over transmitted are wires. of pairs two this than more in use and case cables copper 3 category over transmission which for standard, characteristics BASE-T4 specify 100 the and standard BASE-T2 100 the to expansion an -LX is a solution for structured structured for solution a is BASE-LX 1000 the BASE-SX, 1000 the to addition n bit/s or 100 or Mbit/s M bit/s baseband transmission with a maximum segment length of 400 m 400 of length segment maximum a with transmission baseband Mbit/s m 100 of length segment maximum a with transmission baseband Mbit/s A E T-RJ or VF-45 connectors. The various cable standards will be discussed in a in discussed be will standardsT-RJ cable connectors.or various VF-45 The thernet standard and is used when devices need to communicate with either with communicateto need devices when used is and standard thernet uto- N bit/s baseband transmission via long wavelength.that long means via This transmission baseband Mbit/s wavelength.that means Thisshort via transmission baseband Mbit/s bit/s baseband transmission with a maximum segment length of 100 m 100 of length segment maximum a with transmission baseband Mbit/s egotiation- is based on the on based is Mbit/s. ANP The P ) was created within the framework of the of framework the within createdwas rotocol(ANP) C opper twisted pair cables of various standards are used as used are standards various of cables pair twisted opper CSM C opper twisted pair cables of various standards are standards various of cables pair twisted opper A/CD CSM connectors are commonly used, but used, commonly are connectors SC . A/CD 802.3u refers to a fast a refersto IEEE802.3u 802.z refers to a Gigabit a refersto IEEE802.z 802.z refers to a Gigabit a refersto IEEE802.z I odr o ec 1000 reach to order n 0.a rfr t a Gigabit a to refers IEEE802.3ab connectors are generally used. generally are connectors SC . N way protocolfromway E i/, 250 Mbit/s, thernet with thernet E E National thernet thernet thernet thernet M E thernet thernet T- Mbit/s 353

compendium automatically sets the highest possible speed that can be used for communication In this case, the secondary ring is activated and creates a new and complete between the connected partners. The type of operation is also selected, either ring with the primary ring. However, to do this, the affected stations that are half-duplex or full-duplex, as described in the chapter on transmission methods. directly connected to the ring have to be class A stations. In this manner, failsafe The ANP is optional for 10 BASE-T and 100 BASE-T. It cannot be used with 100 operation is provided by the double ring structure and reconfiguration during a BASE-FX, since interoperability cannot be guaranteed during optical transmission, disruption. FDDI is especially suited for use in MANs, since the maximum distance due to the different wave lengths. The ANP is required with the 1000 BASE-T. between two FDDI stations is 2 km with multimode fibres and 10 km with mono ANP problems can develop if the stations do not react to the sent control packets mode fibres, the maximum distance that can be bridged is 100 km. Although and the half-duplex operation is automatically set. If the communication partner FDDI was originally created for fibre-optics, there is now a copper-based solution should now be manually set to full-duplex, the connection cannot be opened. called CDDI (Copper Distributed Data Interface). With CDDI you can only reach a typical distance of 100 m between the individual stations. 15.6 Ring Topologies 15.7 Protocols/Standards In the ring topologies, the data is serially transmitted in one direction and usually sent from station to station sequentially. There is generally one station Every station that is connected to the network needs both the appropriate that takes over controlling in the ring. This structure will automatically cause network driver and also a protocol driver, which creates the connections to the the entire network to fail if one station fails, which can be avoided in practice higher protocol layers. LAN protocols cover an entire family of protocols and not by using concentrators or double ring structures. In ring topologies the time for just one individual protocol. A protocol is necessary for communication between transmission increases proportionally to the addition of more stations, which, on two instances and it presents, in principle, the mutual language from both the one hand, results in a response time that can be calculated, but, on the other instances. Several important LAN protocols will be presented below. hand, often only allows a lower transmission speed. 15.7.1 Protocols 15.6.1 Token Ring 15.7.1.1 NetBIOS A token ring LAN according to IEEE802.5 places a logical ring topology on a physical star topology. This means that the failure of one station does not lead NetBIOS is an acronym for Network Basic Input Output System and was to the failure of the entire network. The concentrator in such a structure is called developed by IBM in 1984. This protocol is hardware-independent and can thus multistation access unit (MAU). Transmission rates of 4 Mbit/s can be reached in be used in both Ethernet networks and Token Ring networks. It implements the first token ring, but, in contrast, rates of 16 Mbit/s are possible in the second parts of layer 2 and makes functions from levels 2 to 5 available. Layer 3 is only ring. Token ring I works according to token passing, token ring II according to implemented as a zero layer and so this protocol is not routing-capable. Back the early token procedure. Token ring has this name due to a specific bit pattern, then, routing was not so decisive as it is now. NetBIOS allows communication via the so-called token, which goes around in a ring. Only the station that has symbolic names instead of networks, but the number is limited. the free token can transmit. The station that wants to transmit goes from the listening mode to send mode once it has received the free token. The data to be 15.7.1.2 NetBEUI transmitted is attached to the free token, which is then occupied. This means that no other station can change its status while the token is occupied and they stay in NetBEUI is an acronym for Network Bios Extended User Interface and was listening mode. The station that is in send mode transmits until the transmission developed by IBM in 1985. It is an expanded user interface, which is based on is complete or a specific time span has been reached; set response times can be NetBIOS. That is the reason why this protocol is also not routing-capable, since, as kept here. Once the transmission is finished, the station goes back to listening with its predecessor, layer 3 is only implemented as a zero layer. NetBEUI provides mode and releases the token into the ring, so that other stations can transmit. high perfomance in local networks with a maximum of 200 stations and is set up The maximum frame length in token ring I is 4500 bytes, the maximum number as the standard protocol in many operating systems. If communication beyond of stations per segment is 260 and the maximum expanse of cabling per segment layer 2 is necessary, then other protocols come into play, such as TCP/IP, which is is 150 m for UTP and 250 m for STP cabling. In comparison, the maximum described in more detail in the WAN chapter. frame length in token ring II is 16000 bytes, the maximum number of stations per segment is 188 and the maximum expanse of cabling per segment is 100 15.7.1.3 NDIS m for UTP and 150 m for STP cabling. There are, however, various proprietary manufacturer-specific standards that must be observed, also the types of cables NDIS is an acronym for Network Driver Interface Specification Support and is a are a decisive factor. A typical error often made with token ring LANs is including programming interface between layer 2 and the protocols based on it. Here NDIS stations with different speeds. If this happens, the station does not receive a works as a protocol-independent interface that functions as an agent betweent token that is valid for it. This causes a station to miss a beacon, referred to as the network driver and the protocols from other levels. beaconing, and disrupts the network. In addition to the token ring, there is also the token bus standard according to IEEE802.4, where a logical ring topology is 15.7.1.4 ODI placed on a physical bus topology. ODI is an acronym for Open Data Link Interface that was developed by Novell 15.6.2 FDDI and ist suitable for NetWare protocol stacks. ODI works on level 2 and simulates a virtual network card, which allows various packet formats to be sent. ODI is FDDI stands for Fibre Distributed Data Interface and is an ISO Norm that has hardware-independent and is the interface with protocols on higher levels. been generated from ANSI X3T9.5. FDDI is a standard for LANs and MANs and is based on a LWL double ring structure. It can reach data transmission rates 15.7.2 Standards of 100 Mbit/s. There are two types of stations that can be connected, class A stations and class B stations, also referred to as DAS (Dual Attached Stations) and A standard is understood to be a universally valid definition in answer to a set of SAS (Single Attached Station). Class A stations have two interfaces and can be facts. The standard has passed through an approved standardisation procedure redundantly connected to the LWL double ring, whereas class B stations have one and, in the end, forms a technically mature solution. The LAN standards from the interface. There is a primary and secondary ring in the standard configuration, IEEE (Institute of Electrical and Electronical Engineers), an American committee which ensures that the data transmission continues if a cable or station fails. for the creation of standards by engineers, which provides internationally valid

354 transtec = ttec in the Netherlands and Belgium ring networks. ring STP to comparison in reconfigurationtimes Spanning Tree seconds. 60 toThe 30 from be can reconfiguration,which for needed time long the is STP the of disadvantage A sections. remaining the of section best the find to order in algorithm, STP an used. not are sections other all and stations, two between used is section efficient most the only BPDU using other each with communicate bridges the thattraffic.controls on STP Theinfluence lasting a have can which occur,may loops differentways, bridge of variety a in connected are segments and network a in bridges several hardwarechapter.usually arethe there Sincein I themselves. network the analyse they since bridges,transparent called also are branches. individual bridges toThe limited is but network, entire the over out spread not is trafficlocated. receiverthe is waythe This of address target the network the in branch which on deciding and traffic the analysing by domains collisionseparate Bridges loops. bridge so-called create not does and bridge one over goes only it that so communication the regulatesprotocol (STP) tree I 15.7.2.3 Spanning Tree according to 802.1d efficiently.more bandwidth available the use to able be to order in Behaviour) Services). besides possibility service of class another provides which classes. various the to assigned been have which tags, special of means by frames resourcesbetter.use to way simple a offers and 2 level principle. effort best the accordingto sent then are that packets Service). processing.and queuing during considerationinto taken then is priority,which own its has class transfer,file, e-mail, voice,video,etc.as such classes, service various into C 15.7.2.2 C device. the on effect no has this network, local the within change structures network specific VLAN. a to assigned be can device the on station tagging,every VLAN support that used being are components as soon as because flexibility,moreconsiderably has administratorVLAN. port-based the Thus,a with case the physicalis not as VLANs,and logical are they thatVLANs. means This specific to them assign to order in tags, special with stations individual the by the mark and 2 level on tagging,workalso VLAN company.a within accordingto VLANs departments as such differentunits,together group to used usually are VLANs network. entire the an within traffic limit to possible it switches.multilayermake or routersVLANs be can components These 3. level in work that used be tohave components active then necessary, is VLAN the VLAN. within partners with communicate only E an in domains broadcast up set to youallows and LAN for stands Virtual VLAN 15.7.2.1 VLAN according to IEEE802.1q LAN/M for industry.accordingto standards The the to contribution valuable very a networks.make standardslocal Thesefor below.standardsfor explained responsible be is standards,will Work802 group ndividual hardware components in local networks are described in more detail more in described are networks local in hardwarecomponents ndividual spanning the network, local a in used being are bridges two than more f can a VLAN in other.are thateach fromStationsseparated are that thernet oS stands for stands oS I n addition to addition n C oS according to IEEE802.1p iffServ works in a different manner and analyses the analyses and differentmanner a in works DiffServ bridging and management are of special interest.special of are management and bridging AN oS cannot assign bandwidths, which is possible with QoS, but prioritises prioritises but QoS, with possible is which bandwidths, assign cannot oS I f there is a failure in a section, the network is reconfiguredusing is network the section, a in failure a is there f C P rotocol)accordingto lass I t is important, however,up mix important, to is not t o f C S oS, this technology is also used by used ToSService), also of is (Type technology this oS, E ervice and is a way to divide the traffic within a LAN a within traffic the divide to way a is and ervice thernet, in order to increase performance throughout performance increase to order in thernet, (Rapid (Rapid RSTP with problem this solved has IEEE 802.1, which cover extensive standards coverextensive which IEEE802.1, 802.1w, which has substantially shorter shorter substantially has IEEE802.1w,which s (Bridge (Bridge s , but which cannot be used in token in used be cannot which but , E thernet frames, that are sent are that frames, thernet I f communication outside a outside communication f P rotocol C C oS marks the marks oS oS with QoS (Quality of (Quality QoS with oS I f, for example, the f,example,for iffServ (Differentiated DiffServ 802.1q and IEEE802.1q Data transtec = ttec in the the in ttec = transtec HB (P PHB C oS works on works oS nits) and that and Units) E ort Hop ort thernet thernet E very very IT

N etherlands and Belgium and etherlands faulty signals or collisions to a specific segment, which leads to an overallincrease an to leads which segment, specific a to collisions or signals faulty limit can it i.e. detection, error for capacity certain a has repeater the addition, progression.signal during cycle and level regeneratesthe it that means which functions, regenerationmentioned,already as performs, repeaterThe that. beyond extended be should and reached been has segment a of expanse physical maximum the if used is repeater A functions. regeneration performs an in 1 layer on works that component active an is repeater A 15.8.1 Repeaters 15.8 Hardware -Active C aggregation. link term the in networks mobile trunked regional describe to used already are terms these but past, the in used often were trunking 10-Gigabit feasible economically Gigabit 100 based fibre-optic- and copper- with used technology.is aggregation network Link complete the modify to having without rates data higher attain to technology existing use to way simple a is aggregationthrough bandwidths higher get overall.reducedto is possibility bandwith The the only wherebymaintained, be can connection connection. The a of failurecomplete a to lead not does partners communication two between link a of failure operation.the way This failsafeguarantee also and bandwidth the increase to order in link, logical one just into links physicalseveral together group to method a is aggregation Link 15.7.2.4 LinkAggregation according to IEEE802.3ad transparency found here, with regard to the connected stations in a network. network. This a in stations connected the toregard here, with found transparency the refersto also standstill. name a Thisto communication data bring can loops transmission. such data forTransparentsince essential, pathsare various bridges are there if and used being are bridges several if arise can Loops network. a in path communication loop-free a find to STP the Transparentuse bridges bridges.transparent as known are they then protocol(STP), tree spanning the own. their on network local the of addresses the learn they since bridges, learning as segment.known arethis These just to limited is communication then segment, same the in locatedreceiver are the and sender network. entire the to traffic the expand to necessary is it whether decides O their of basis the on analysed are packets data and other each fromseparated are domains collision bridge a of help the Furthermore,with functions. detection error has also and functions that. beyond extended be should and reached been has segment a of expansephysical maximum the if used are here.requiredBridges is intellegencemore bit a since ring,e.g. tokenused, aredifferent topologies LL the in bridges an in 2 layer on works that component active an is bridge A 15.8.3 Bridge backplane. the form which busses, internal its by determined largely is hub a of used. areperformance Thethey where on depend also hubs of extendability hubs.and sizeenterprise The e.g.or workgroup,used, are departmental, they where location differentiatedthe areaccordingto Hubs media. transmission and different connectors using segments and stations various connect can and an in 1 layer on works that component active an and repeater multiport a basically is hub A 15.8.2 Hub topology.the of limits physical the beyond segment the extend to used only is and transparent protocolcompletely is it 1, layer on worksrepeater a Since performance. in ver time, a bridge learns which stations are located in which segments and segments which in located are stations which learns bridge a time,ver sublayer.however,C LL work, the often on devices The E thernet links is especially important to be able to close the gap up to an to up gap the close to able be to important especially is links thernet E bit/s and 1000 and Mbit/s s usually work on the on work usually LANs thernet E . The hub is the concentrator in a star topology star a concentratorin the is LAN. hub The thernet Mbit/s omponents M A addresses.C E thernet network. The terms trunking and port port and trunking terms network. The thernet E thernet networks, whereby aggregation of aggregationwherebynetworks, thernet nglish. This is why the work group selected selected groupwork the why English. is This M

A sublayer and very seldom on seldom very and sublayer C I f a learning bridge uses bridge learning a f I t takes over regenerationover takes t E sublayer when sublayer C thernet LAN. thernet E and LAN thernet I n M I f the f ost 355

compendium is based on the fact that the bridges do not change the packets and only work on create routing tables that depict the constantly changing network structures. A layer 2 in a LAN. Besides these types, there are additional types of bridges, such as layer 3 switch that is not equipped with the appropriate routing protocol is, if you encapsulation bridges, translation bridges and source routing bridges. prefer to see it this way, a device without a leader. In addition to packet switching The encapsulation bridge encapsules the frame of a protocol in the data area and route processing, the intelligent network services make up a large part of the of another protocol, in order to transmit data over sections where this type of advantages in a layer 3 switch. They are so-called value added services, such as protocol is not supported. In this manner, an Ethernet frame, for example, can be management, security, IP multicasting, and (QoS). transmitted over an FDDI network. This process is often referred to as tunneling. The quality of service aspect is especially emphasised in this context. QoS covers A disadvantage of this process is that a protocol overhead develops due to this all processes that have such an influence on data traffic, that the service arrives at nesting, and bandwidth is then wasted on the transmission medium. Translation the receiving end with a set quality level. The fact that QoS plays an ever growing bridges translate the addresses from one type of frame standard into another role can be traced back to the type of data that has to be transmitted. In the past, frame standard. In order for communication to take place, the frame types must file transfer and e-mail were the primary types of data, but, in the meantime, be adjusted in terms of minimum and maximum lengths, so that both standards applications from the voice and video sectors have become more and more can work with these frames. Source routing bridges are used to facilitate important. These applications require specific service qualities in terms of latency, transmission between different token ring networks. rate of packet loss and bandwidth. There many different solutions from institutes for standardisation on how to implement QoS. At the moment, there are very few 15.8.4 Switch QoS models that are uniform across all manufacturers, and quite often a proprietary solution from a specific manufacturer is preferred, which limits A switch is basically a multiport bridge and an active component that works on flexibility in terms of network equipment. Using intelligent network services, the layer 2 in an Ethernet LAN. Switches perform, as the name implies, switching overall performance of a network can be further increased, independent from functions and are used as the concentrator in networks with star topologies. The additional increases in the bandwidths. Due to this and all the reasons described core of a switch is the switching fabric, which determines the performance. The above, layer 3 switches will become more prevalent in the future. forwarding rate and filtering rate also contribute to the performance of a switch. The forwarding rate describes the maximum number of data packets that can be Cascading/Stacking forwarded per second whereas the filtering rate indicates the maximum number The terms cascading and stacking are often used in conjunction with solutions in of processed data packets and MAC addresses that can be administered. Switches the switching sector. Both terms describe an option of connecting one or more are additionally divided according to the switching process into cut-through, switches to one unit, but in different ways. The basic difference between fragement-free and store-and-forward switches. With cut-through switches, the cascading and stacking is that in cascading the switches are connected by data packets are simply sent through to keep the switch delay time as short as standard ports and by special stacking ports in stacking. In cascading this can lead possible. After the target address has been determined, the cut-through switch to the interface between both devices becoming a bottle neck in the does not wait for the transmission of the data to be completed, but immediately configuration which has unfavourable effects on network performance. Stacking, forwards the next data. Other than the target address, no other information in in comparison, uses ports that are provided especially to connect several switches. the packet is necessary to complete this process, which can dramatically reduce Due to this, both devices can be connected to each other with the complete the delay time. In the fragement-free process, the switch waits for a specific time bandwidth and form a unit. The stacking solutions from the manufacturers make period in order to avoid collisions in the target segment. The most common it possible to communicate with the physical switch stack as a logical switch via an process is the store-and-forward process where the data in the switch is buffered IP address. This makes it much easier to do maintenance work on the device. You before it is forwarded. The advantage of this is that the data goes through an should, however, note that stacking solutions are manufacturer- and product error detection and error correction process, whereas the disadvantage is that a specific. Other than classic stacking, many manufacturers provide solutions where longer delay time develops. you can administer many switches that are dispersed across the network in just one stack, even if they are not physically connected. But this is a borderline case Layer 3 Switching to classic stacking. The term layer 3 switching and other synonyms for this technology are being used more and more. This paragraph should clarify this technology a bit more. There is Modular Switches an entire group of synonyms, other than layer 3 switching, which are used for this In terms of hardware, you can differentiate between two basic types of switches. technology. This is why terms such as switching routers, layer 4 switching and One of them are fixed configuration switches, that are set by their port structure multilayer switching are also used. However, all of these terms describe the same and are primarily used in the edge area. The other type are modular switches, technology, but the term layer 3 switching will be used below. It is a mistake to which have a freely configurable port structre and are mostly used in the core believe that a network’s performance can be increased purely by the bandwidth or area. the packet switching performance. Many of the switches that are used in the edge areas of local networks can be An advantage of layer 3 switches is that the decisive components, besides packet expanded to modular switches. These switches have a fixed port configuration switching, are the route processing and intelligent network services. These three with additional extendable module slots, that can be equipped with various components are the foundation on which a layer 3 switch is built. A layer 3 switch interface modules. Since a variety of modules are usually on offer, the installation is a high-performance router that is especially used on local networks. In contrast can be carried out flexibly with Ethernet standards and media interfaces. There are to a classic layer 2 switch, it also works on the higher layers in the OSI reference two common types of modules. One is an uplink module that is usually only model. The physical implementation also differs from that of classic routers. Thus suitable for a specific device from a specific manufacturer. Additionally, network with classic routers, packet switching is based on microprocessor-based engines, manufacturers are making every effort to develop a uniform module standard. but layer 3 switches are based on ASIC hardware-based technologies The result of these efforts is the GBIC, which stands for Gigabit Interface (ASIC=application specific integrated circuits). For this reason, classic routers work Converter. GBIC modules are Gigabit Ethernet modules, that can basically be in the Kpps range and layer 3 switches in the Mpps range. integrated into all GBIC module slots. However, different manufacturer-specific Besides these different implementations, classic routers and layer 3 switches have developments have led them to a point where general interoperability cannot be a lot in common in one respect. Using route processing, the hardware-based guaranteed, in spite of uniform module standards and form factors. Even edge packet switching of a layer 3 switch is turned in the right direction. Route switches can be flexibly connected with these modules. processing is a software-based implementation that should, however, be regarded Modular switches that can be found in the core area, can, in contrast, usually be as independent from packet switching. Protocols such as RIP, RIPv2, OSPF, OSPFv2 modularly configured. Consisting of a chassis, switch controller slots and interface are used with route processing, which control the packet and route and then slots, a variety of port configurations can be implemented. Due to the central

356 transtec = ttec in the Netherlands and Belgium However, the disadvantages are that they are sensitive to electromagnetic to sensitive are they However,that are disadvantages the installation. inexpensive and flexible, fast, are cabling copper-based of benefits C 15.9.3 C effort. of lot a with accomplished be only can changes structural that is problem Another distributor.floor the to connected are that stations of number the accordingto decreases also terminal the at available is disadvantage.that a bandwidth The is times delay about bring components,which active However,of number the cabling.copper inexpensivemore with section horizontal the in used be also can structure.decentral the to due fail-safemore are sections the that is this distributor.of advantagebuilding An central the to floors the run and section horizontal the bundle distributors floor a forconcept The failure.of point single a createsAdditionally, structure connected. this be tohave that devices of number the to correspond to enough least at ports, manyhave connected. individually be to has station every required,since is cabling of lot A disadvantages. some has concept However,costs.installation and this even effort less with created be can network redundant a and necessary as not are rooms distributor with structures building centralising,costly this section. With connected the of bandwidth complete the to accesstheory, in has, station connected every and times delay cause could that components active more no are thereconcentrator. that are this of Advantages this with connected directly are stations concentrator;all powerful one on the of concept The stations. connected the to flows data the of all which over network a of backbone the literally is cabling. backbone structured The of development the from arisen haveconcepts backbone Two important C 15.9.2 Backbone here.used is section horizontal term section the in all, of Last this. describing when used also is section vertical term the why is which distributors, floor the to connected are distributors building bridged. be tohave that distances the to due technology,cable fibre-optic-based uses generally section distributors. primary The building the at end or begin that lines cable redundant of means byprovided section primary I possible. as differentapplicationsmany as transmit to able be tohave possible technology. cabling in developed been has methods cabling universalfor model structure A 15.9.1 StructuredC network. high-performance a forfoundation the are they since components,passive the selecting when accountinto taken be must demands standards.future highest with The use for suitable be also and investmentsecurity cabling. the of cost the as well as implemented, be tohave that speeds transmission the bridged, be to has that distance the all,above and, expansion for interference,options to immunity are cabling for criteria Selection changing.also are medium transmissionphysical the on changing,demands constantly is technology network Sincecabling. network the about talking are they technology, network with conjunction in mentioned are components passive When 15.9 Hardware -P network. the of failure total preventa to redundant normally are components existing the switches, such of function cabling is often used in the tertiary and secondary sections. sections. secondary and tertiary the in used often is cabling copper-basedwhy bridged. is be This can that distances shorter and influences mplementation of this is carried out by creating various steps in a hierarchy.a in steps various creatingThe by out carried is this of mplementation opper cables are widely used and many and used widely are cables opper the floor distributors are connected to the terminals, which is why the why is which terminals, the to connected are distributors floor the opper-Based C I n this case, structured means that as few transmission media as media transmission few as that means structured case, this n is cabling that is not limited to just one building, which is often often is which building, one just to limited not is that cabling is distributed backbone distributed oncepts abling collapsed backbone collapsed assive C abling omponents P

lus much less effort is needed for cabling and it and cabling for needed is effort less much lus D ue to this, the central concentrator also has to has concentratorcentralalso the this, to ue

compresses all of the floor backbones backbones floor the of all compresses is worked out differently.out manyworked Here is E thernet standards require them. requirestandardsThe thernet I

n the n secondary section secondary C I abling must provide must abling n practice, two practice, n transtec = ttec in the the in ttec = transtec

the tertiary tertiary

N etherlands and Belgium and etherlands copper cables are suitable for the for suitable are cables copper 600 of bandwidth a with 7 category 250 of bandwidth 100 of bandwidth a with 5 16 of bandwidth a with 3 category are categories common section. The transmission a in characteristics electric all of requirementsminimum the define classes classes. performance Theseperformance various into divided are cables Furthermore,copperoutside. cable entire the as S outside. the on shielded not is cable entire the but for shielded. is outside cable entire the but shielded, not are inside pairs S/UTP shielded. is cable entire the nor inside pairs wire individual the neither where other.each from distinguised are cables of types following the shielding,categories. standardisedthe toregardWith and shielding of type differentiatedthe areaccordingto Twistedcables pair transmission. forwires of pairs tworequire services all wires.Almost eight copper-based lines. twisted The of pairs more or one with cable copper a fordesignation the is connector.RJ-45 a have and TwistedBASE-Tused pair are10 standard the since copper-basednewer shield.all forouter Twistedcables an pair coveredby then is this of All line. aluminium or copper a by as well as insulation, by surrounded is that follows.line coreas a up is Thereset is cable (Bayonet BNC with cable RG58 thin BASE-2 10 A with cable RG8 thick uses BASE-5 C cable. pair twisted the and cable coaxial the used, widely are media transmission 50/125 micron fibres, or 5000 m with 9/125 micron fibres. A variety of connectors connectors of variety fibres.A micron 9/125 with m 5000 fibres,or micron 50/125 and 62.5/125 using bridged be can m 550 fibres.BASE-LX, micron 1000 125 With 50/ using bridges be can m 500-550 or fibres micron 62.5/125 using by bridged used. is that medium transmission the the on depending bridged, be can sections transmission of variety A micron. 9/125 of diametercore a have only fibres mode single contrast, micron. 50/125 and 62.5/125 fibres: multimode with used are that diameters common jacket. two areand Therecore fibre-optic the of diameter the is optics fibre- differentiatingforbetween criterion decisive A distances.larger bridge to possible it makes thus and attenuationlower in results which line,straight almost an in spread can light fibre,the mode single a of diametercore small very the to mode. one in transmitted is today.light fibres,the primarily mode single With light. the of spreading wave-shaped in results which parable, a like shaped is that index refractive a has strand gradient jacket. and core the between occurs light the of reflection total a that so jacket, fibre-optic the than index refractive higher a has core fibre-optic fibre,the index step modes.multimode several the Within transmitted is light fibres.fibres,the monomode multimode Withor single the and differentiated.fibres are multimode the are Thesefibre-optics of Two types index. refractive their is jacket the and core the in fibre-optics cladding.the glass differencebetween The a by surrounded is core,which the in transmittedare signals light Thefibre-optics. the protect that buffers primary and secondary into divided jacket plastic a of composed is cable sections. fibre-optic The primary and secondary the in used generally LWLis passive cabling and coppercomponents. why active expensiveis This the as well as install, to difficult and flexible very not are cables the that are distances.large bridge to ability the as well rates,as transmission high influences, electromagnetic to insensitivity are cables fibre-optic of Benefits many and used widely are cables Fibre-optic 15.9.4 Fibre- E for suitable are 7, and 6 categories from cables especially and cables, copper the for suitable are cables thernet standard 1000 BASE-T.1000 standard thernet hielded -5 and 10 BASE-2 10 and BASE-5 10 the for used are cables oaxial S hielded stands for stands T wisted ptic-Based C Optic-Based T wisted E thernet standards have four twisted pairs of wires or a total of total a or wires of pairs twisted fourhave standards thernet S P creened / creened Hz according to the accordingto MHz air where the individual wire pairs inside are shielded as well as shielded are inside pairs wire individual the where air P air where the individual wire pairs inside are shielded are inside pairs wire individual the where air E abling Hz accordingto MHz thernet standard 100 BASE-T,100 standardenhanced 5 thernet category U nshielded E (AttachmentUI thernet standard 10 BASE-T,10 standardcopper 5 thernet category Hz accordingto MHz U Hz accordingto MHz M T -SX, 220-275 m can be can m 220-275 BASE-SX, 1000 sing UTP raft Draft ultimode gradient strands are used are strands gradient ultimode wisted stands for stands E 50288-2-1, category 6 with a with 6 category 50288-2-1, EN n the other hand, the multimode the hand, other the On I thernet standards require them. requirestandards thernet Neil nternational Standard July 2002, July Standardnternational P air, where the individual wire air,individual the where C Unit S/STP oncelman) connectors. oncelman) The 50288-4-1. EN U I nterface) connectors and connectors nterface) stands for stands E nshielded 50173, category category 50173, EN thernet standards.10 thernet E thernet standard and standard thernet E thernet standards thernet Disadvantages or TeflonPVC T S wisted creened / creened C STP ategory 3 ategory stands

P air I n D 357 ue

compendium are used, such as ST, SC, MT-RJ or VF-45 connectors, which are not dependent on evaluate them. There are now three different versions of SNMP, which are the type of optic fibre used. constantly readjusted to fit current demands. There is also a second version of MIB, MIB-II, which supports a variety of new protocols that deliver even more 15.10 Management detailed information. Due to the object identifier, MIB-II is downward compatible to MIB-I. We also have to mention RMON, which stands for Remote Monitoring. Network management is an essential tool for network administrators to control RMON is a MIB that was developed to enable proactive network management. complex local networks. The simple network management protocol (SNMP) is Information has to be called up regularly in MIB, which could lead to an increase decisive to accomplish this. SNMP was developed in 1988 and is the foundation in the network load. With RMON, the information is collected and saved by the for administration in TCP/IP networks; it is also based on the application-oriented device itself. You can also set limits with RMON, so that a message is only sent layers in the OSI reference model. The goals during the development of SNMP if this limit is exceeded. This also reduces the network load considerably. Due were software and hardware independence, as well the mimimum use of to the fact that RMON has to analyse every frame, delay times, which have a system resources. An SNMP environment is made up of a managment station, negative influence on the network performance, may occur. RMON1 works on on which the corresponding SNMP management software is installed, and the the levels 1 and 2 and RMON2 works on levels 3 to 7 in the OSI reference model. network components that are equipped with the appropriate SNMP agents. The RMON2 does not supersede RMON1, but is an expansion to achieve transparency SNMP agents rely on the management information base (MIB), in which the for higher protocol layers. All of these components make it possible for the characteristics of the network components are described in order to guarantee administrator to manage local networks, which is essential in complex structures. that the management station only requests data that the agent can deliver. However, you must note that the use of management components can also lead The main tasks of the management are monitoring, controlling, diagnosis and to delays in the network and pay attention to the physical principle that every problem analysis. The management station requests the data from the SNMP reading changes the current status. agents in order to convert them appropriately so that the administrator can

358 transtec = ttec in the Netherlands and Belgium 16. in a in FDDI, the to Thanks offered. be to is service a if employedAT be as to such technology,has technology M a transport a not is but connections, logical and lines providingfor technology a merely is SDH Since users. the between connections logical establishing for and cabling logical for technology a networks,town of basis the formstoday SDH purposes. a in up set be can networks local users. connected all forrouting economical or transmission as such services value-added providing in I universities.for networks campus as or facilities, their kilometres. 2 over least a viewpoint, geographicFrom a services. providing for used is and operator an by up set is which networks, other against delimited a standpoint, specific the from Regarded area. larger a forexploited be can LANs in inherentadvantages the a form to LANs these connecting By LANs. individual generally networks, a viewpoint, conceptualFrom the (M called are (WANs) networks wide-area and (LANs) networks local-area between are breadth,geographic of terms in which, Networks 16.1 D n addition to classical services of data transmission, the task of a of task the transmission, data of services classical to addition n ANs). M is realised with very high quality and, above all, shorter transmission shorter all,above and, quality high very with realised is AN efinition M

etropolitan Area

high-speed technologies used, service provision ATservice or used,technologies DQB high-speed M M s are typically set up for conurbations, towns and townsconurbations,for up set typically are ANs M M M extends as far as 150 kilometres, but at but kilometres, 150 as far as extends AN for cross-location and cross-facility and cross-locationfor AN is a network, interconnecting individual interconnecting network, a is AN M telephony,IP is in turn an individual network, network, individual an turn in is AN N M etworks - etropolitan

image and video and image IP I n addition, virtual virtual addition, n transtec = ttec in the the in ttec = transtec A rea M

consists AN N etworks etworks M M , all AN, A N s N etherlands and Belgium and etherlands therefore discussed in the respective chapters.respective the in thereforediscussed and WALANs are in and used Ns also are standardscorresponding the with together SDH and G.703 ATas and well technologies,FDDI as Thetransmission M throughsystems transmission the by proposed are standards G.709 and G.708 G.707, the 139 or 34 of speed the with G.703 connection. write or read as used is case each in bus one as connections, duplex full supports bus double the levellowest the At communication. of direction the on based is bus the of buses. selection two The the of one on information sending by nodes other with communicates and buses implemented. be can topology ring and star Both directions. opposite in transmit buses unidirectional instance.forof language, concept coded The of transmission isochronous and asynchronous both permits method This a of backbone the as and kilometres 100 to up of distances for used be can and standardisation group working the by defined was DQB D 16.2.1 Distributed Queue 16.2 Transmission Technologies for UpaM Setting 10-Gigabit the of regions speed high- the into venturingmeanwhile is technology a WAin further than A time N. E thernet. thernet. M AN.

M ach node in a in node Each ANs: ual Bus(DQB) E thernet, thanks to its latest stage of development: of stage latest its to thanks thernet, DQB. H with 155 with SDH or Mbit/s B ring structure is connected to both to connected is structure ring DQB

802.6 in the framework of framework the in 802.6 IEEE B is that two optical fibre optical two that is DQB

AN bit/s accordingto Mbit/s as standardized as IEEE M AN 359

compendium 17. Wide Area Networks - WANs

17.1 Definition than in IPv4. Apart from the increased address space and the attendant possibility of assigning Geographically dispersed networks are called Wide Area Networks (WANs) an address that is unique worldwide to every host, a decisive advantage of and provide communication solutions for organisations or private parties. The IPv6 addressing is that so-called site multihoming is possible. Thanks to site geographic range can extend to 1,000 or even 10,000 kilometres. multihoming, multiple IPv6 addresses can be assigned to hosts, allowing a host to The basic technology employed is either data packet switching or circuit use the services of multiple providers. In addition, IPv6 addressing offers a defined switching. WANs often consist of multiple individual networks interconnected, header format, making the processing of the addressed packets far more efficient. for example, through ISDN, X.25 or ATM. Secure transmission in WANs is today What is more, IPv6 also offers the option of using security headers. realised through Virtual Private Networks (VPN). An example of a WAN without which life cannot be imagined is the analogue 17.3 Protocols in a WAN network in use for decades for telephony. In this network data is transmitted at a transmission rate of 56 Kbit/s. Due to increased demands on transmission rates in WAN protocols are protocol families at Layer 3 of the ISO/OSI reference model. The WANs due to new services such as video and voice via IP, transmission rates of up Layer 3 protocols most widespread today will be introduced below, including the to 622 Mbit/s are meanwhile realised with the use of ATM. With the development transport and routing protocols. of 10 Gigabit Ethernet, the migration into even faster transmission rates is just round the corner. 17.3.1 The Internet Protocol (IP) In Europe WANs are usually operated by telecommunication providers. The task of the platform-independent Internet Protocol (IP Protocol) is the 17.2 Addressing in a WAN transfer of data across multiple networks from a sender to the receiver. IP is at Layer 3 and is responsible for regulating the transfer of data fragments forwarded Addressing in WANs is based on the Internet Protocol (IP) addresses assigned to from Layer 4 to Layer 3. An IP packet is defined as having a minimum size of the active network components connected to the WAN via the netid and to the 576 bytes and a maximum length of 65,535 bytes. Due to this variability and the hosts via the hostid. IP addresses are unique and in IP version 4 consist of 32 bits variously defined transmission structure and packet length per subnetwork to be subdivided into four octets for simpler handling. traversed, an IP packet may be split in the course of the transmission path. Such Two IP addresses are however excluded from this addressing scheme: splitting is called fragmentation. Reassembling refers to putting the fragments the IP address in which all the bits of the host ID are set to 0 and the address in back to restore the original state. which all the bits are set to 1. The first one serves as the general description of a The structure of an IP data packet (also referred to as datagram) always remains network address, the last one is a reserved broadcast address. basically the same regardless of the fragmentation. The Network Information Centre (NIC) manages and distributes IP addresses. The main constitutents of an IP packet are in consecutive order: The header with The NIC allocates the netids, the hostids are allocated by the respective owner of the information on the packet proper, the sender and receiver address and all a netid. the information serving for transfer control and troubleshooting. The header is followed by optional information such as security rules for the data transfer. The

Distinction is made between three classes of addresses: data information to be transferred is associated to the last bits. Class A addresses can address 128 networks, because the first octet forms the Transfer of IP packets is packet-oriented, connectionless and not guaranteed. In network address but the first bit in this octet is 0. The other three octets are for the event of a buffer overrun along the transmission path, an IP packet is rejected assignment to hosts, which corresponds to the theoretical possibility of assigning when arriving at the overloaded node. However, thanks to the incorporated unique addresses to 16,777,214 hosts each. frequency check sum and the directly built up communication between sender and receiver, retransmission takes place in the event of an error. Since IP packets are not

Class B addresses can address 16,384 networks with 65,534 hosts each. The first regarded as an aggregate of information units, but independently of one another, two bits in the first octet of IP addresses in Class B networks are defined by 10 and their transfer is arbitrary, i.e. connectionless. Each data packet takes its optimal thereby characterise Class B networks as such. Generally the first two octets form path from the sender to the receiver. The TCP protocol based on the IP protocol is the netid, the last two octets form the hostid. responsible for sorting the IP packets into the correct order at the receiver.

Class C addresses can address 2,097,152 different networks with up to 254 hosts 17.3.2 The Transmission Control Protocol (TCP) each. In Class C networks the first three bits of the first octet indicate 110 for unique identification. Only the fourth octet is used for assigning IP addresses to The Transmission Control Protocol (TCP) is at Layer 4 of the OSI reference model hosts, the other three octets serve for assigning netids. and is responsible for processing the data stream from Layers 5-7 in fragments capable of being processed for the IP protocol at Layer 3. TCP divides the flood

Due to the explosive development of the Internet, freely available addresses of data, for instance of a file to be transferred, into packets by virtue of the according to IPv4 are meanwhile running low and therefore a unique IP address is prescribed rules and transfers them to the transmission protocol. almost always assigned only to the active network components directly connected to the WAN. All the other devices connected behind the respective WAN interface The TCP/IP (Transmission Control Protocol/Internet Protocol) protocol stack receive addresses from the Private Name Space which, however, are then no longer defined by the U.S. Department of Defense is available on every major computer unique worldwide. To prevent any misrouting, the IP addresses are identified as platform. It was not designed for a special information transport system, such as belonging to the connected network at the interface to the WAN via Network a LAN protocol, but rather for use across different media and computers. Hence, Address Translation (NAT). TCP/IP is the most suitable protocol for networking heterogeneous systems. It is capable of connecting computers running under Unix (or any of its variants such

To counter the problem of insufficient addresses according to IPv4, a new address as SunOS, Digital Unix, HP-UX, AIX), OpenVMS, DOS, or Windows. A range of structure has been developed. applications have been based upon TCP/IP, which are usually included under the In the Internet Protocol version 6 (IPv6), compatible with IPv4, the addressing is title of the protocol: e.g. ftp for file transfer,telnet and rlogin for remote control based on the hexadecimal system. An address according to IPv6 is 128 bits long or remote login, electronic mail, Web browsers, etc. and consists of a series of eight 16-bit long parts, providing four times more bits

360 transtec = ttec in the Netherlands and Belgium (variable-length subnet masking) or manual or masking) subnet (variable-length M VLS support not does R IP network. the throughoutinformation routing the all advertises protocol,routinewhich structure. in simple very is operandi modus its protocoland routing oldest the is route.RIP traversinga in delay or bandwidth as such criteria, metric further heed not does RIP next. the to component routing one from so-called on purely based is which of metric The 17.3.4.1 RIP protocoltypes. both featuresof the combines EIGRP protocols.vector distance on based those convergentmorethan significantlyprotocols are state link on based discovery, networks its after directly network the in change topology a advertising routers. By and links forinformation routing of comparison little very signifying table,routing new entire an as not change, single a as advertised is information routing new updates,i.e. triggered support hand other the protocolson state Link protocols. routers.RIP the on excessiveload and excessivetrafficto procedure leads this tables routing complex highly information. new the with full in tables routing own their replacing by information this process who intervals, periodic at neighbours direct their to tables routing full their send rumour”“routingand by principle the accordingtooperateservice. They convenientin operation and installation simple scalability,howeverbyaccompanied limited convergencethe and comparativelyslow their by characterised procotolsare vector Distance protocol.state link or vector distance a is it whether protocolsis routing featurebetween distinguishing general second A protocol.to specific advertised information the of basis the on selected is network the throughroute best The the i.e.information, route its advertises network. the in neighbours their from changes of changes. protocolshear such Theseto reaction in tables routing their update and network the in recognisechanges administrator.the for tables routing updating on input involvingless of virtue by use this forgrowingnetworks.predestined are They D table.routing the in manually entered high. very is tables routing updating manually resourcesfor on burden the because connections, dial-up for protocolsrouting Static protocols.routing dynamic and static between made is Distinction protocol.protocolto from decision. forwarding its takes router a which upon based value the as defined protocolis routing a of it. metric from Thederived metric the protocoland routing the through up set table routing the a on WAin based is N router a of decision forwarding network. The destination a and sender a between packets of routing and sequence the forresponsible protocols arerouting The P 17.3.4 Routing transfer, lostpacketsare notre-sent. the transfer hasbeensuccessful. UDPhence doesnotguarantee reliability ofthe UDP sendsdata to thereceiver, butwithoutreceiving that acknowledgement ofsendingdatagramsthe possibility direct.UDPoperation isconnectionless. is responsible purely andwas developed for to thetransport give applications model and, like T The 17.3.3 U 15 hops. 15 is RIP with operating network a for distance summarisation. routemaximum The ynamic routing protocolsrouting ynamic U R outing ser ser Datagram P D I set up on the basis of a dynamic routing protocol, each routerprotocol, routingeach dynamic a of basis the a on WAn up set N

atagram I nformation rotocols CP , is a transport protocol, isatransport ontheIPP baseddirectly P rotocol ( rotocol (UDP) are generally used only in relatively small networks or networks relativelysmall in only used generally are P

rotocol( on the other hand are often to be found in larger,in found be to often are hand other the on E very new route and every change in a route has to be to has route a in change every and route new very UDP F and OSPF

, RIP , ) is, like T RIP v2 and v2 ) is a distance vector routing protocol,routingthe vector distance a is )

addresses it knows, to its neighbours. its to knows, it addresses IP S are examples of link state protocols.state link of examples are IS IS- CP are examples of distance vector distance of examples are IGRP hop counts hop , at Layer 4oftheOSIreference I n large networks having networks large n D . A hop is the transfer the is hop A . ynamic routing protocolsrouting ynamic M etric calculation varies calculation etric I t is a flat, classful flat, a is t transtec = ttec in the the in ttec = transtec

rotocol. UDP

N etherlands and Belgium and etherlands throughput and throughput delay,bandwidth, reliability,calculation: metric forparametersfollowing the network. the of overalltopology recordingthe route,but best the identifying only not tables with more,routers is Whatnetwork. a of topology the in changes unscheduled of advertising immediate the updates,i.e. triggered protocols.state link and Thus vector distance of E of weaknesses The networks. small in only used is convergenceand time added. are link a to time delay the and bandwidth minimum that in complex more is calculation protocol.routing vector distance proprietary a is IGRP count. hop the on only based still is criterion metric the and hops 15 max. RIP in persist still RIP Twoof weaknesses exchanged. informationrouting the is then only and requested is connection a frozenuntil I connections. point-to-point with operating networks small in implemented mainly RIP network. a of topology the in changes unscheduled updates,i.e. triggered advertise also can protocols,it vector distance of typical updates periodical RIP this singlecommunication. to the onlyforThe isreserved connection theparties duration ofthecommunication since thisone setupcircuit canbeusedonlyfor bandwidth isnotfullyutilized by the transfer, thebandwidthislostfor the communicationfor between sendersandreceivers. every I bandwidth ofacircuit, since aseparate circuit issetupacross thehierarchies in thehierarchy. I approach, thoselower inthehierarchy are associated linearly to thosehigher I 17.4.1 LineSwitching switching. WANs transfer data eitheronthebasisofcircuit switching ordata packet 17.4 Data Switching ina WAN routers.the on load the minimising thus and relativelysmall kept be to tables routing the allowing permit routers,that border area called are areas various the between backbone. routersThe the via conducted be to has non-backbones between traffic routing All non-backbone. as connected directly be tohave areas other the all which to 0, area as defined be to has backbone environments,a OSPF I transfer.the in node last and first the between provide connections the bandwidths of sum the and costs path the on based is route best the determining for calculation metric the OSPF manual and VLSM supports convergence functions, acceptable.not are subject is RIP which to limitations routing O 17.3.4.4 OSPF on based D route.best the determining for 17.3.4.3 IGRP/EI 17.3.4.2 RIP n circuit switching, theswitching points are arranged according to ahierarchical using requirementwhen basic a is network the of knowledge n-depth t is also used for dial-in connections, because in RIP in because connections, dial-in for used also is t nhanced ue to the proprietary, both proprietary, the to ue pen v2 , the , and supports VLSM. supports and RIP of version extended an is S hortest hortest IP I C v4. nterior I nterior isco hardware.isco

v2 F is optimised for use in large, scalable networks in which the which in networks scalable large, in use for optimised is OSPF M P anual ath M D n such networks datan suchnetworks canbetransferred withtheentire G G TU ue to the pure distance vector functionality, it has pretty slow pretty has it functionality, vector distance pure the to ue ateway were eliminated in the extended version, extended the in eliminatedwere IGRP ateway F irst ( irst . However, only bandwidth and delay are used as standard as used are delayHowever, and . bandwidth only address summarisation is also supported. supported. also is summarisation address IP

OSPF Only R R outing and IGRP outing ) is a link state protocol and was developed fordeveloped was protocoland state link a is )

C isco hardware supports these protocols.these hardwaresupports isco P P rotocol, was developed by developedrotocol, was rotocol. can be employed only in networks networks in only employed be can EIGRP v2 though: the hop count is limited to limited is count hop the though: v2 combines the characteristics characteristics the combines EIGRP and VLSM supports EIGRP C support have routinghave support EIGRP ompared to RIP to ompared v2 the routing table may be may table routing the v2 address summarisation. address IP

F offers enhanced offers OSPF I n addition to the to addition n f, however, the C isco and hence and isco , its metric metric its , provides EIGRP EIGRP summaries summaries IP OSPF. , I v2 is v2 n 361 I n

compendium communication (pre-allocation). as possible to guarantee the best return on investment. The carrier of the leased Advantages of circuit switching are that the transfer is almost entirely delay-free, line establishes the connection either via a physical connection or via an assigned there is full protocol transparency and uncomplicated routing with little burden on channel with the use of either frequency modulation or time multiplexing. Leased computing resources. lines generally enable synchronous data transmission. The very long time required for making and dropping the connection has negative repercussions. Further, no redundant paths are provided and the poor utilisation of 17.6.2 Circuit-Switched Networks the bandwidth is to be regarded as a drawback of circuit-switching. Circuit-switched networks provide a dedicated connection between sender 17.4.2 Data Packet Switching and receiver for the duration of the transmission. Thus, for example, analog connections across the telephone network (PSTN) run on circuit-switched Data packet switching splits the information (data) to be transferred into networks. For the duration of the connection the line is not available for any other chunks, so-called packets. In single packet switching the sender allocates clear connection, even if the available bandwidth is not fully utilised. Further examples addressing, send sequence, flow control and error correction characteristics to of circuit-switched connections are ISDN and asynchronous, serial connections. the packets to enable the receiver to evaluate the entire information transfer and to establish whether and when transfer was fully concluded and how the overall 17.6.3 Packet and Cell-Switched Networks information is to be composed from the individual packets. The individual packets are forwarded from the sender via various nodes to the receiver in a technology In packet- and cell-switched networks, the carrier creates permanent virtual circuits based on buffers. Each node stores the packets, checks them and then forwards (PVCs) or switched virtual circuits (SVCs). In these networks the users share the the checked packets to the next node. This method of receiving, buffering, resources available and use various paths through the network for their data checking and forwarding is also called Store and Forward. transmission. This enables the carrier to improve the rate of capacity utilisation of The transfer time of data-switching technologies is significantly shorter than that its infrastructure compared to point-to-point connections and thereby to optimise of circuit-switching technologies, particularly due to the shorter time for making the return on investment for all involved. It is employed in networks that are and dropping connections. What is more, the bandwidth utilisation is optimised intended to use X.25, Frame Relay or SMDS (Switched Multimegabit Data Service). and redundant paths can be used. A drawback of the technology is that it is more ATM is a technology used only in pure cell-switched networks. ATM divides data error-prone. into cells of defined size and transfers them via a physical medium with the use of digital signal technology. ATM cell transfer can be either asynchronous, queued or 17.5 Transmission Methods in a WAN with the use of a multiplexing technique.

17.5.1 Synchronous Transmission Methods 17.6.4 Digital Subscribe Line (DSL)

Synchronous transmission methods transfer data within prescribed time intervals DSL is a transmission technology enabling high transmission rates through the synchronised using a clock pulse. Each connected end station reads out the clock traditional telephone cabling. The four versions of DSL are Asymmetric Digital pulse in the header of the data packet and subordinates itself to it. Synchronous Subscriber Line (ADSL), High Data Rate Digital Subscriber Line (HDSL), Single-Line transmission methods are faster and more efficient than the less ordered, Digital Subscriber Line (SDSL) and Very high Data Rate Digital Subscriber Line asynchronous transmission methods. (VDSL). Due to the fact that DSL and telephone transmission can operate on the basis of different frequencies, both data streams can be transferred in parallel. The 17.5.2 Asynchronous Transmission Methods most common DSL version is ADSL. ATM and DSL will be described in further detail in a section below. Compared with synchronous transmission methods, asynchronous transmission methods do not have a prescribed clock pulse. The data packets of asynchronous transmission methods have start and stop bits for error recognition. A transmission 17.7 Data Transmission in a WAN error occurs if the difference between a start and a stop bit upon arrival at the receiver is longer than it was at the sender. 17.7.1 Analogue Data Transmission in WANs

17.6 WAN Technologies Probably, the most extensive and familiar WAN is the public telephone network, in which dial-up connections are established to provide mainly voice communication In general WAN technologies are based on the lower three layers of the OSI links. The use of this network as a platform for data transmissions via modem or reference model: the Physical Layer, the Data Link Layer and the Network Layer. As facsimile will continue to be significant. a rule the services are provided via one of the following technologies: - by leased lines (dedicated lines) 17.7.2 Digital Data Transmission in a WAN - by circuit switching - by packet switching 17.7.2.1 Integrated Services Digital Network - ISDN - by cell switching ISDN combines the use of voice, text, images and data into one single network. All the switching techniques mentioned are based on virtual channels. The This means that only a single ISDN connection is needed for access to a variety of channel to which a data packet belongs has to be capable of being read out by telecommunications services such as telephone, fax, file transfer utilities, etc. The the service provider on the basis of the information in the packet. high-speed, error-free data transmission, quick connection and growing popularity make ISDN not only attractive for newcomers to data communications, but also 17.6.1 Leased Lines for users who up to now have been dependent on analogue modems. ISDN operates through the already installed telephone lines for the end device. The main Dedicated lines are point-to-point connections permanently reserved for data difference with respect to the current telephone network is that signal transmission transmissions. They are not just set up when a transmission has to be made, but is entirely in digital form, right up to the end device. Data communications are established permanently regardless of whether or not data is flowing. For between central relay stations and local relay stations is already completely in leased lines it is therefore advisable to optimise the bandwidth utilisation as far digital form.

362 transtec = ttec in the Netherlands and Belgium errors inthefaxtransmission. However circumstances thismethodiscomplicated can result andundercertain in for this card. This converts digital software data for theanalogue faxdevice. cards, whichdonothave amodemchip, iftheappropriate isavailable faxsoftware modem. I chip oradigital signal processor, allowing thedevice to beusedlikeananalogue ISDN cards andISDNterminal adapters feature anintegrated analoguemodem them. distant endsonlywork iftheISDNenddevice supports This iswhy several data communication isconsiderably more complicated. C example viaananaloguemodemoraccess to then any othernetwork, I 17.7.2.1.4 C Windows NT, for instance, alsouses C is aconsiderable improvement. C standard. The more recent C PC-based ISDNsolutions. P two versions ofC Telekom were at theforefront ofthe development ofC C also supports C supporting ISDN enddevices (e.g. ISDNcard). I the aimofcreating andthe between application auniformsoftware interface C 17.7.2.1.3 C canbeusedbothwithandwithoutacontrolpermanent connections channel. maximum of30Bchannels(PRI)are available perconnection. These dedicated Digital also possible(analoguetelephone system, mobilephones, X.25). the dial-upprocedure to any ISDNsubscriber. to othernetsare Dial-upconnections For the The ISDNsystem includesthefollowingtypes: connection distinct 17.7.2.1.2 C installations. consequently, wherever shouldbeusedfor possible, ofconnection new thistype M ISDN, whichare onlyavailable onanational basis. features are service data isto orlinetypes neededin betransmitted orifparticular protocols usedat therespective locations. However, problems may when arise telephone, noproblems shouldbeencountered ifthere isadifference inDchannel has grown to 30worldwide, 20ofwhichbeing inE non-E astheE known is theE channel protocols for ISDNexist: Oneisthenational standard 1TR6andtheother over theDchannel. This isthereason why, inGermany for example, two D standards were developed for theprotocols usedfor sendingcontrol information connection-related technical data. Following theintroduction ofISDN,several desired (e.g. oftheservice indication ofthetype voice orfax)aswell asother caller‘s number, whichthenappearsonthedisplay ofanISDNtelephone, an is transmitted over aseparate channel(Dchannel). This information includesthe I 17.7.2.1.1 C ISDN card oradigital faxdevice (faxgroup 4),orthere is ISDN access to anetwork I 17.7.2.1.5 C f thedistant endisbeingaccessed for viatheanalogue telephone network, between two ISDNdevicesf aconnection ismade, firstofall control information f there isadigital enddevice at the distant endsuchas, for example, aPCwithan API standsfor C eanwhile, E uropean countries. The providers numberofnetwork offering E uropean standardised protocol DSS1orE-SS1. The E-DSSIstandard is fixed lines ISDN dial-upconnection t isalsopossibleto communicate withanaloguefaxdevices usingISDN ommunication viaISDNwithAnalogue Distant Ends onnection Types ompatibility API ommunication viaISDNwithDigital Distant Ends API to beused together withany which ISDNapplication software ISDN hasbecomeuro- thestandard inmany countries, and ISDN protocol,uro- even thoughnowadays itisalsousedinmany . German companies fromAPI. German D andthe theISDN sector ommon ISDNApplication I API inuse. C provide channels. permanentlyswitched digitalA 64-kbit/s ractically allISDNcards this ractically for PCinstallation support API version 2.0provides cross-platform which support, API versions 1.0and1.1have beendesigned for

, aBchannelisprovided by for anISDNservice ross-platform support isalsothereasonross-platform why support n principal, C 2.0 in its Remote Access Services (RAS). API 2.0initsRemote (RAS). Access Services nterface andwas developednterface with API enablesany ISDNcard urope. When usingonlythe API. There are at present onnections withanalogue onnections transtec = ttec in the the in ttec = transtec ISDN uro- eutsche N etherlands and Belgium and etherlands forms thequadruplemultiplex ofthelower-order interface higher-order interface. Further E-X are definedthanks interfaces to amultiplex hierarchy inwhichthe perinterface. 64 Kbit/s This basicbitrate isalsocalledtheE0interface. The E superseded innewinstallations by thesynchronous digital hierarchy (SDH). in G.702ofITU-T PDHhasmeanwhile andisalsocalledtheE0interface. been various bitrates allbasedonthebitrate of64Kbit/s. This basicbitrate isdefined asynchronous. PDHwas standardisedat that timeandhas in1972byCCI the TT transmission between nodesthat donothave identical clockspeeds, hence are The plesiochronous digital hierarchy (PDH)definesadigital technology for 17.7.2.2.1 P 17.7.2.2 Digital M achieved onlyby utilizinghardwarebothstandards. supporting Absolute inISDNcommunication flexibility withdigital distant endscanbe defined inthestandard V.110 for 2,400 to 38,400bit/s. the difference between 64,000bit/sandthe required bit rate. This technique is This iscalledbitrate adaptation. For that purpose,to makeup fillbitsare inserted communications at 64,000 bit/s,supporting thenthedata rate hasto beadapted. rate is64,000bit/sperISDNBchannel.I The X.75protocol usesthefullbandwidthofISDN,andthusdata transmission V.110 standard. is usuallyexchanged viatheISDNBchannelincompliance witheithertheX.75or provideror to (Datex-J-Service, aservice C throughput of34,368 MbpS. E-3: Aroute over whichfour E2linesare combined to form alinewithdata throughput of8,448MbpS. E-2: Aroute over whichfour E1linesare combined to form aline withadata interfaces: Thanks to multiplexing, thefollowing increases are possibleasvariations ofE channelsfor voiceservice anddata transmission à64Kbit/s. transmits or2Mbit/s. upto 2,048Kbit/s areThese 2,048Kbit/s madeupof32E0 PDH multiplex hierarchy inE (alsoCEPT1)istheP The E1transmission interface one linewithoutany inthroughput. reduction As aresult ofthemultiplexing, thedata ofmultiplelinescanbesent onviaonly One thenspeaksofmultiplexing. To increase thebandwidth32E0linescanbecombined to form E1 interface. an 17.7.2.2.3 PDHM for transmissions. I here all32available data channels(alsotimeslots)are 64Kbit/s individuallyused versionG.704 isthestructured ofG.703. G.704operates likeG.703only exactly G.704 and theU isthemostcommoncable. withtheexception Balanced service oftheNetherlands either through balanced pair)orunbalanced (120ohmtwisted (dual75ohmcoax) communication equipment suchasrouters or multiplexers. Transmission iseffected to form adata channelwith2Mbit/s. G.703ismostlyusedfor connecting that all32available datameans inthisconnection channelsare 64Kbit/s combined G.703, butthistransmission rate Unstructured canalsobeutilized onlyinpart. transmissions withthemaximumtransmission rate of2Mbit/spossiblewith dataunstructured streams. The G.703isasaruleusedfor unstructured rates of64k/bitsupwards. I The G.703ITUrecommendation usedfrom transmission aninterface describes G.703 17.7.2.2.2 PDHStandards uropean digital multiplex hierarchy PDHoperates onthebasisofabitrate of

.K. lesiochronous Digital Hierarchy (PDH) ultiplexing ultiplex Hierarchies n G.704 time slot 0 is reserved forn G.704timeslot0isreserved thesynchronisation though. t is a bit-oriented digital interface for digital transmitting interface t isabit-oriented urope andthereby thefirststageofmultiplexing. I f thedistant endisnotcapableof ompuServe, I ompuServe, rimary Rate I Rate rimary nternet P nterface (PRI)ofthe nterface rovider), thendata 363 t

compendium E-4: A route over which four E3 lines are combined to form a line with a data 17.7.2.2.5 Synchronous Digital Hierarchy (SDH) throughput of 139,264 MbpS. E-5: A route over which four E4 lines are combined to form a line with a data SDH, or the American version SONET, is a flexible transparent multiplex structure throughput of 565,148 MbpS. and is characterised by its high availability and reliability. Transmission rates between 155 Mbit/s up to 40 Gbit/s can be realised through SDH. In comparison Plesiochronous is equivalent to almost synchronised. This is indicative of the fact with PDH, SDH provides the advantages of bundling lower transmission rates (e.g. that in PDH certain unavoidable differences in clock speed can occur in the various 2 Mbit/s) into a higher transmission frame. For the information transmission SDH multiplex stages, which are made up for by filler bits for compensation in the uses STM signals (Synchronous Transport Module) transmitted in the STM basic upper hierarchy stages. data frame STM-1 at 155 Mbit/s. The higher transmission rates can be realised by virtue of multiplexing. The clock pulse in SDH, compared to the clock pulse in PDH, 17.7.2.2.4 Multiplex Techniques is controlled by a uniformly, centrally generated network clock pulse. Accordingly all the hierarchy stages have the same clock rate and lower hierarchy stages can There are five different types of multiplexing: space division, frequency division, be directly accessed without having to pass through all the multiplex stages. This time division, code division, wavelength time division. direct access is realised through STM transport units.

17.7.2.2.4.1 Space Division 17.7.2.2.5.1 SDH Multiplexing

With space division multiplexing (SDM) the physical transmission media are The STM basic data frame STM-1 allows for a data transmission rate of 155 Mbit/s. combined into one frequency band or cable. In relation to digital transmission, An increase of the data rate is always possible by quadrupling the lower multiplex space division multiplexing is used to combine several ISDN connections at one hierarchy. This results in the following data transmission rates: local exchange in a primary rate interface. STM-4 with 622 Mbit/s transmission rate as the second hierarchy step STM-16 with 2.5 Gbit/s transmission rate as the third hierarchy step 17.7.2.2.4.2 Frequency Division STM-64 with 10 Gbit/s transmission rate as the fourth hierarchy step STM-256 with 40 Gbit/s transmission rate as the fifth hierarchy step With frequency division multiplexing (FDM), a broad frequency band is divided into several smaller frequency bands that can all transfer data parallel, but independent 17.7.3 Serial Interfaces in a WAN from each other. FDM is used, for example, for telegraphic connections. FDM is not commonly used today, since it has been superseded by time division multiplexing Serial interfaces are often used for dedicated line connections in WANs. The most common ones are described below. 17.7.2.2.4.3 Time Division 17.7.3.1 V.24 With time division multiplexing (TDM), the individual transmission requirements are processed one after the other. Every channel is assigned a time slot to transfer its V.24, developed by ITU, is an interface transmitting 20 Kbit/s over 15 metres. data. If the time slot is exceeded, transmission is disrupted and the next channel It is probably the most widespread interface technology between Data is given the right to transfer for the length of its time slot. However, in TDM you Terminal Equipment (DTE) and data circuit-terminating equipment, generally a must differentiate between synchronous time division multiplexing (STD) and telecommunications connection (e.g. a modem). V.24 is independent from the type asynchronous time division multiplexing (ATD). of transmission and supports both syncronous and asynchronous transmissions. The physical interface is the 25-pin miniature D-sub connector. The synchronous time division multiplexing (STD) technique defines transmission frames, of which each are made up of a specific number of time slots with fixed 17.7.3.2 RS-232 sizes. Every user is assigned a specific time slot within the transmission frame, and during this period it can send or receive data. RS-232 is the serial, bidirectional interface, which is based on the functional characteristics of V.24 and transmits asynchronous data. RS-232 defines serial The asynchronous time division multiplexing (ATD) technique transmits the transmission in both the half-duplex and in the full-duplex mode. RS-232 is data asynchronously in information units that have a fixed or variable length. specified for two serial channels which can both be active independently of each Information unit assignment takes place with channel identification numbers, other and can send and receive at one and the same time (full duplex). so-called channel identifiers, that are included with every packet. ATD is thus also called address multiplexing. This procedure is also called cell switching if data In practice RS-232, with the full name RS-232-C for the current version, is used packets with fixed lengths are used during transmission, and data packet switching for the communication of a computer with another computer or another serial if the packet lengths are variable. device. Since the computer internally uses a parallel data stream, RS-232 specifies how the Universal Asynchronous Receiver/Transmitter (UART) chip at the Terminal 17.7.2.2.4.4 Code Division Equipment (DTE) interface of the computer converts these parallel data streams into a serial data stream, which is then sent serially, one bit after the other, to Code division multiplexing (CDM) allows you to simultaneously transmit data from another computer. As the DTE of the computer, the UART also communicates several senders over one medium (frequency or channel). This is made possible with other serial devices which then have the counterpart to the DTE as the by individual coding of the data using a bit pattern per sender, which enables the communication interface: the Data Communication Equipment (DCE) which, for receiver to reproduce the information using the code. CDM is used in UMTS. example, modems and routers use.

17.7.2.2.4.5 Wavelength Time Division 17.7.3.3 X.21

Wavelength division multiplexing (WDM) is used for transmission over fibre-optic X.21 is the ITU-T standard for access to digital networks. It applies to all packet- cables. In WDM, various fibre-optic wave lengths are transmitted paralled as a light switching and circuit-switching networks, but also for dedicated lines, with digital current. Simultaneous transmission can even be done in full-duplex mode. access. X.21 applies to both symmetric and asymmetric signal lines. Throughput rates of up to 10 Mbit/s can be achieved with X.21, as a rule however 2 Mbit/s

364 transtec = ttec in the Netherlands and Belgium conferencing have andvideodistribution beenincreasingly incorporated into werenetworks setupseparately. Since multimedia applications suchasvideo few years ago, thisproblem would nothave since arisen voice, videoanddata applications suchasvoice andvideotransmission. A suitable for time-critical Although Frame Relay offers clearadvantages over X.25,this technology isless which conforms to theE3standard. the router isX.21. orbridge The highestspeedclassat present achieves 34Mbit/s, at 2Mbit/s(E1).I category withthenext kbit/s is used. operates interface at aminimumtransmission speedof64 The smallentry of D can beemployed invariouspublicorprivate networks, e.g., inthe Datex-M Service being usedto itsfullpotential. D economic advantages wherever isnot ofapoint-to-point connection thecapacity the data throughput ina WAN. I compared withX.25. Therefore, Frame goodchoice for Relay isavery improving tocorrection upperprotocol layers. C known to problems X.25is Thehave efficiency are henceerror evaded byshifting devices. ofdataguarantee packets, delivery butleaves thecorrect thistaskto theterminal operates onlevels 1and2oftheISO/model. Thus, theprotocol doesnot its comparatively low protocol overhead. I Frame Relay isatransmission system similartoby X.25,whichischaracterised 17.7.4.2 Frame Relay Telecom. in Germany, Telepac inSwitzerland, or DatanetISTEL,British Netherlands 1inthe There are separate data for networks data transmission through X.25,e.g. Datex- P Bit/s are possible. second and thegoodtransmission quality. Transmission speedsof300to 64,000 capable remote routers. The callset-up timeofone mainadvantages are theshort withX.25- localnetworks suitablefor linking are particularly X.25 packetservices circuits. circuits orswitched virtual permanent virtual and are dividedbetween max.16logical channels. These are connections either I Access P Service Network between thethird layer layer above andthetransport itintheOSImodeliscalled for thehigherapplication layers are defined for X.25transmissions. This interface andtheterminal system theconnection 3, establishingandclearing connections control anderror schemesfor recognition checking ofthetransmission. I physical andtransmission proper connection takeplace, inlayer and 2thesecurity oriented to thethree lowest layers oftheOSIreference model. I format for X.25transmissions into consideration. The ITU-T recommendation is I toand connected apublicdata network. thesynchronousdescribes operation ofaterminal operating inthepacketmode X.25 The probably standard best-known for packet-mode transmission techniques is 17.7.4.1 X.25 17.7.4 X.25andFrame Relay networks. Several telephone linescanbecombined with V.35. data andaccess networks between packet-oriented equipment tointerface these V.35 operates withbandwidthsgreater andislikewiseusedasan than19.2Kbit/s 17.7.3.4 V.35 for X.21. interface dedicated linesare setupthrough isthestandard X.21.A15-pinconnector essential whichhave innetworks until now onlybeenusedfor data transmission. LAN and WAN networks, theintegration ofaudio andvideoinformation isnow n X.25 up to 4,096 virtual connections are connections setupbetween thelinkedterminals n X.25upto 4,096virtual eutsche Telekom. For access network the . X.25was developed by ITU-T. I oint (NSAP). n comparison withfixed lines, FrameRelay offers epending onthecountry, Frame Relay technology n itsrecommendation for X.25,theITU-T onsiderably higherdata rates are achieved t alsotakesthetimingandtransmission n contrast to X.25,Frame Relay only SNI n terms of the standard interface ton terms ofthestandard interface ( S ubscriber ubscriber N transtec = ttec in the the in ttec = transtec etwork etwork n X.25layer 1the I nterface) nterface) n layer N etherlands and Belgium and etherlands which reaches thegigabit range. managersare Network pressing theirsuppliers for multimediaapplications. The keyto ATM isthetransmission rate, however, time behaviour. ATM offers this feature, and is therefore a technology intended the hardware components andsoftware involved real- mustpossessapredictable result. Therefore, ifmultimediadata are to beintegrated all into thenetwork, oftheimagesequence istheunavoidable andinvideoinformation, jerkiness click, E data, suchasaudioandmoving pictures. more orless. inthecaseoftime-critical As arule, itdoesnotmakemuchdifference whetherafile transfer takesasecond 17.7.5.1 Data Transmission inAT M Networks networks. ATM hashowever alsofound itsway into somelocalnetworks. packet-mode transmission technology developed mainlyfor useinwidearea A 17.7.5 Asynchronous Transfer M guaranteed are characteristics notprovided. clausesshouldalsobeagreedchecked andcontractually bindingpenalty onincase a Frame Relay provider, oftheoffered thequality shouldbecarefully network or for Round Trip D forusual inthismarket CIRs( providers differ significantly, even more vary buttheperformances so. I Various providers are offering FrameRelay connections. The prices quoted by these ATM, plays arole here. The packet-switching transmission method-theAsynchronous Transfer M ATM recognises ( permanentconnections resources elements ofthenetwork canbeusedefficiently for several connections. determined path through whentheyare thenetwork generated. I I no permanentchannelsare establishedbetween theterminal devices involved. on whichpacketswitching suchas X.25are networks based. U links, thesameprinciple The transmission ofvirtual itselfisbasedontheprinciple adapter. adaptation mechanism into severalpartitioned ATM cells andreconstituted at theirdestination usingthe in asingleAT M cell, thedifferent layers lengthpacketsofhighernetwork are issuitable for As mostinformation allservices. cannotbeaccommodatedin theory ATM forum hasfound asalomoniccompromise withthecell size for AT M, which suitable for voice transmissions, longblocksfor data andvideotransmission. The cell format usedby ATM, suchproblems are avoided. blocksare moreVery short cause problems, e.g. otherapplicationsfiletransfers. during blocking D data flows for individualapplications. With other topologies, longpacketscan means that guaranteed bandwidthscanbeassigned ifthere are several competing of allcells, there isacalculabledelay inthetransmission ofany information, which ofATcharacteristics M result from thisdata structure. Thanks to theuniform length for userdata, 5bytes are for reserved checkinformation. essential Some of fixed length(53 bytes), whichare called cells (cell relay). 48 bytes are available I protocols unnecessary. seamlessly. This makesadditionalgateways transposing from LANto WAN range. AT M isthefirstnetwork technology that can integrate LANsand WANs But thisisnowhere neartheupperlimit:AT speedsinthemulti-gigabit M permits generation ofAT 155 Mbit/s or622perport. supports M switches typically provision ofthebandwidththat happensto beneeded. The present-day I andlookforstudy themarket migration paths that leadtheuserto ATM. technology tofrom aneveryday solution.C beingahigh-end backbone. This iswhere AT M represents apromising investment, andwill shift and system integrators to provide solutionsto theirbandwidthproblems inthe nstead, the cells of a particular connection are transported alongapreviously are connection nstead, transported the cells ofa particular n AT ofinformationM, alltypes (audio, videoanddata) are conveyed inpackets n additionto itsspeed, there ofAT isthescalability M, meaningtheflexible ven aslight delay intheaudiodata stream willcausean audiblediscontinuity or synchronous T ransfer elays to bewithinareasonable range. Before decidingupon M SAR ode ( ( C S egmentation ommitted I ATM ode (ATM) ) is a fast, service-oriented, asynchronous, ) isafast, service-oriented, nformation provided to Rates) beactually PVC a nd

s, P R ermanent eassembly) ofthesendingAT M V irtual irtual ompanies should sing thisprinciple, n thisway, the C ircuits) aswell ue to the t isnot ode, 365

compendium as switched connections (SVCs, Switched Virtual Circuits). These in turn can consist process. For this to happen, the added costs per port of ATM networks will either of virtual point-to-point or point-to-multipoint connections. have to drop dramatically. However, for applications such as high performance workstations, servers, image processing or virtual reality, ATM is already an Connections can be established with a specific grade of service,QoS (Quality interesting alternative to other topologies. of Service). Each device connected to the ATM can be allocated the required But how can existing applications be integrated into an ATM network? The easiest bandwidth statically or dynamically, by means of the PVCs or SVCs mentioned, way would be to exchange the existing computer‘s LAN interface for an ATM this being limited only by the capacity of the ATM hardware. After the link has board and install the ATM driver. Unfortunately, things are not that easy. ATM lacks been established, one can be sure of being able to use the bandwidth of the one property that conventional LANs offer: support for broadcasts, multicasts, transmission channel requested alone, without being disturbed or even interrupted and the use of MAC addresses. These are basic prerequisites for the use of most by other stations waiting to transmit. network protocols, such as TCP/IP or IPX.

The system attempts to meet the respective demands in the best way possible. The solution agreed on by the ATM Forum to this dilemma is called LAN Emulation In principle, ATM offers three different categories of service: guaranteed services, LANE 2.0. In a network combining ATM with standard LAN technologies, LANE predictable services and services that utilise an available bit rate to the best consists of four software components. The LAN Emulation Services (LESs), possible effect. A guaranteed connection is needed for supporting constant bit consisting of three components, are made available to one or more computers rates (CBR, Constant Bit Rate) for which all cells must reliably transfer data to their connected to the ATM network as ATM applications or as an integrated part of an destination with only a slight time deviation, as is the case for speech information, ATM switch. The LAN-Emulation-Client (LEC) is installed as the fourth component for example. But there are also applications with variable bit rates, for example file on the end devices. transfers and e-mail, for which a service that utilises the available bit rates as best LAN Emulation Services: possible, suffices. Here, there are no hard-and-fast guarantees. Rather, the cells The LAN Emulation ConfirmationS erver LECS co-ordinates all LESs and LECs. are simply transmitted as best as possible on the bandwidth that is left available by The LAN Emulation Server LES translates MAC addresses into ATM addresses and connections with a higher priority. In the ATM Forum‘s specification, these services vice-versa. are listed under the term Traffic Management. The Broadcast and Unknown Server BUS continues to forward broadcast and 17.7.5.2 Interfaces multicast packets using ATM point-to-multipoint connections and forwards packets to unknown computers, until the destination address is found. At the beginning A large number of ATM transmission interfaces are defined both for wide area of a transmission, the LEC contacts the LECS using a uniform network address to networks (WAN) and local area networks (LAN). The transmission speeds for the exchange general LANE information. The LECS provides the client with the address physical media range from 1.5 Mbit/s to 622 Mbit/s. The differences between LAN of the corresponding LES. With this, the LEC establishes a connection with the LES. and WAN specifications are primarily in the transmission medium. While single Then, ATM address, MAC address, frame size and LAN type are agreed on. The mode fibres and coaxial cables are primarily provided for the WAN sector, multi- LEC passes broadcasts and packets with unknown destination addresses to the mode fibres and Twisted-Pair cables are in more prevalent use in the LAN sector. BUS which forwards them.

17.7.5.3 Availability and Standardisation The missing MAC level features are then emulated for clients, for whom ATM then appears either as an 802.3 (Ethernet) or 802.5 (Token Ring) network. This The ATM Forum, an interest group of ATM manufacturers, is the driving power enables communication between conventional LANs and ATM. behind the standardisation of ATM. It provides the rather hesitant ITU (successor to CCITT) with proposals. The definition of cells was completed some time ago 17.7.6 Digital Subscribe Line (DSL) as well as the definition of the functionality of the hardware components of ATM systems. Also defined was the way in which the hardware of ATM systems has to DSL is a transmission technology enabling high transmission rates through function, how connections are to be set up, how they should be monitored and the traditional telephone cabling. The four standardised versions of DSL are which logical interfaces are needed. Asymmetric Digital Subscriber Line (ADSL), High Data Rate Digital Subscriber Line ATM also comprises further specifications though. According to the manufacturers, (HDSL), Single-Line Digital Subscriber Line (SDSL) and Very high Data Rate Digital it presents no problem to adapt the products to different standards by means Subscriber Line (VDSL). The term xDSL is often also used when referring generally of firmware upgrade. This ensures compatibility of products from the same to DSL in all its various forms. manufacturer with specifications from the ATM Forum. If ATM products from Due to the fact that DSL and telephone transmission can operate on the basis of more than one manufacturer are to be used together, the products used must be different frequencies, both data streams can be transferred in parallel. compatible. 17.7.6.1 Asymmetric Digital Subscriber Line (ADSL) 17.7.5.4 Migration to Existing Networks ADSL is the most widespread and most frequently used DSL standard. ASDL is an Due to the substantial differences between ATM and previous network asymmetric wide-band data transmission technique using conventional twisted- technologies, flexible integration of ATM products in existing networks is necessary pair copper lines. For ADSL communication an ADSL modem has to be installed at to protect hardware investments. This long-term process of adaptation is called either end of the line, both at the local exchange and at the subscriber‘s end. In migration. The migration of ATM will take place in several steps. ATM switches the ADSL method the copper line is divided into three channels: the downstream can be used in large networks as a collapsed backbone. Previously installed ring channel from the service provider to the end user, the upstream channel in the or bus-type backbone cables are brought together at the ATM LAN Access switch, opposite direction and a channel through which PSTN and ISDN communication which connects all segments and devices of the network in a star-type topology. takes place simultaneously by adding a splitter. Additional ATM switches or ATM routers can be connected to such an ATM switch, The signal to be transmitted is divided into diverse parts transmitted through as well as terminal devices using ATM adapters. Existing network topologies are different carrier frequencies. In addition to the discrete multitone (DMT) method connected to central ATM switches by means of routers with ATM functionality. specified in the ADSL standard, the CAP/QAM method is also used. With ADSL Most of the manufacturers of large network concentrators offer modules for ATM transmissions up to 6 Mbit/s can be realised through normal telephone cable, for connectivity. bidirectional transmission a second frequency band with transmission rates up to Setting up a continuous connection to terminal devices is the last step in this 640 kbit/s is available.

366 transtec = ttec in the Netherlands and Belgium regarded as those to be taken most seriously, because it is very difficult to uncoverto difficult very is it seriously,because most taken be to those as regarded attack. an such of victim the of image the on presentation party‘s a attacked Webon the server,attack altering is example Another lost. are data the all and late too is beforeit attack this notice even not maythem. owner deleteThe or alter to plans,development to access gain to tries attacker an e.g.editing,when or accessfor authority respective the without company.a to greatestdanger the constitutethreats confidentiality or integrity So-called 17.8.1 Attacks and Targets ofAttack like or WebServers servers public that and them, alter possibly resourcesand network the to access gain thus network, companythe internal fromWAinfiltrators the penetrate no can that tapped,N that WAtaken be be cannot communicationto ThereforeN-based has care security.to threats posing as regarded be to also arecompany a in operations smooth disrupting at levelled attacks Willful altogether.them lose to or parties third with profit and turnover its share to having of risk the runs company the parties, third of knowledge the to come data and information respectivecompany. the the Should of possession sole in remain must company‘ssuccess a underpinning implemented.solution unique The be security network that is company a of success the for condition basic A in 17.8 Security WANs range lessthan300metres. between 300and500metres andshort-range VDSL isdividedinto three ranges: long-range between 1and1.5kilometres, mid- 2.3 Mbit/s. pair copper lines. The upstream transmission speedsare between 1.5Mbit/sand hence from nodeto user, of13M bit/s to 52Mbit/sare reached through twisted- definition television transmission. With VDSL,downstream transmission speeds, aggregate bitrates of12,24,48und 72Mbit/s, canhence alsobeusedfor high- having bitrates of8Mbit/s, 16Mbit/sand28bit/s. The classhas symmetrical transmission,and symmetrical theclassfor operation theasymmetrical modes Mbit/s are reached. I With VDSL, dependingonclass, transmission symmetrical bitrates ofupto 72 transmitted over bothcopper andfibre opticlinks. transmitted at highspeedover relativelydistances. short This variant ofDSLcanbe VDSL isusedwherever data orasymmetrical streams symmetrical have to be DSL) 17.7.6.4 DataVeryDigital Line(V Rate Subscriber High km. only onepairisneededfor transmission. The maximumdistance for SDSLis3.5 to 2Mbit/s(E-1line).UnlikeHDSL, transmissionsymmetrical speeds of64kbit/s Like HSDL,SSLisamethodfor modewith transmissions inthefull-duplex 17.7.6.3 Symmetrical Digital Line(SDSL) Subscriber to 5kilometres. With HDSL,transmission rates ofupto 2Mbit/scanachieved over distance ofup Relay. I HDSL isidealfor private branch exchanges andisanalternative to localFrame Data Digital Rate Line(HDSL) Subscriber 17.7.6.2 High kilometres. forkilometres transmission speedsabove 1.5Mbit/sandbelow that upto 6 applications. information-on-demand The distance nodecanbe4 to thenext Examples ofthisare pay-per-channel, pay-per-view, and video-on-demand alike. andinteractive ADSL canbeusedfor videoservices services distribution company‘s entire success.company‘sentire changes.any repair and I nternet. n addition,itcanreplace orE-1lines. amplifier-supported I ntegrity threats are when an attacker tries to alter corporate data corporate alter to tries attacker an when arethreats ntegrity I t is impossible to put a figure to the damage this may to do to the to do to may this damage the to figure a put to impossible is t n terms of classes, distinction is made between asymmetrical n termsismadebetween asymmetrical ofclasses, distinction C ancellation of data may even destroy the basis for the for basis the destroyeven may data of ancellation

M I n many companies integrity threats arethreats integrity companies many n ailServers cannot be attacked. be cannot ailServers

transtec = ttec in the the in ttec = transtec

N etherlands and Belgium and etherlands attacker could, for example, read out passwords and use them for further attacks, further for them use and passwords out readexample, forcould, attacker it. misappropriate and switch backbone a onto data flowing copy could attacker Forthe example, configurations Altering - components network between confidence of basis the on up set services security Attacking - - damage: inflicting operationsfor used be can access illegal this which in ways of multitude router,a a are thereinstance,or forswitch a to access attack. of target the be also can components network Active processed. and accepted be longer no can users actual the from inquiries that router a to fortargets be also up.giveforcedto thus and inquiries or e-mails with flooded example forare servers, or as WWW Webaccess,such servers public with Servers discontinued.even is service the or rendered be longer no can service the that extent an such to recipient the occupy to destination the to sent threat further a constitutes (D service of denial or availability Service such. forwarding and altering office, head the transferredthroughWAthe intercepting data to is branche.g.from a N, kind this of attack further A all. at not difficulty,greator with only ascertained be can found.be can data the to changes noticeable no again, network the left has attacker the once because, identify to difficult morecontents. even aretheir attacks learn These to order in data C network? the to - it? to access what gains who and company - route? what via - example: forbe, might policy a such of matterpolicy. respective subject the The of basis the on taken are measures security Area-specific risks. identified the minimising of objective the process.ongoing an is policy a such of implementation and with. developmentThe dealt are risks these A company.the in handled is it how and danger potential the defines policy security internal An protection. their formeasures appropriate take and them one‘‘sclassify toin attack network, of ownpoints potential the know to is attack outside against protection effectiverequirementfor basic The provider. network every for risk incalculable almost representan described dangers potential the taken, are measures security appropriateany Unless 17.8.2 P unavailable. services rendering or data secret to access write or read unauthorised gaining described: already those as targets same the followattacks availability. network or applicationsThese hosts, on attacks be may latter The attacks. specific further,formore exploited then are which attack, of points and weaknesses the derives he employed.analysisFrom applications this and used topology,services whole the the including network, entire the of picture a get to tries attackerattack. The of targets unprotectedpotential,for combed attack. subsequent specific forinformation gain to hopes therebyattacker the when attackers fortargets become networks Entire the to access that reconfiguredso be couldrouter access authority.An access having as future in him identifies firewall the that way a such in configuration the alter could attacker firewall,the a on user.regular a as password the via network the to access gaining I D D D onfidentiality threats are levelled at gaining access to secret or highly confidential highly or secret to access gaining at levelled arethreats onfidentiality nterceptingdata security policy security efinition of a user policy. What conduct is expected of a user who has access has who user a of expected is policy. user conduct Whata of efinition the to importance of levelhierarchy. what data of a is of data efinition What and whom by when, accessedcontrol. accessis network data Whatof efinition rotective M

documents the risks to which a network is exposed and how and exposed is network a which to risks the documents I D

f a remote access server becomes the victim of an attack, the attack, an of victim the becomes server accessremote a f easures andStrategies oS attacks. For example, the attacker sends to many packets many to sends attacker Forthe attacks.example, oS P

olicies are as a rule broken down to individual areas with areas individual to down broken rule a as are olicies

I n a n D oS attack, a vast amount of data or requests is requests or data of amount vast a attack, oS I f the attacker has copied the secret data, this data, secret the copied has attacker the f I nternet is no longer available.longer no is nternet oS), targeted at making services unusable, services making at targeted oS),

I n this way a network might be might network a way this n I nternet connections can connections nternet

I f an attacker gains attacker an f

I n case of attack of case n

367

compendium A further example: advantage is that the caller (normally the company switchboard) pays the cost. This A company wants to develop rules for Internet access as part of a fully is useful but irrelevant from a security viewpoint. Nevertheless, dial-back systems comprehensive security policy. The security policy says that Internet access is still have disadvantages: available to users only for certain, necessary services, that no user is allowed to install applications himself and then use them, and that all transactions - The device, not the user is authorised contravening the defined security measures are to be fully documented via an - Dial-back seldom works for mobile users audit. As a result of this definition of the rules for Internet access, the company - Authorised users not using a pre-defined device can be locked out implements the following precautionary measures: - The system can be interfered with by call forwarding - The firewalls allow the passage only of HTTP traffic and e-mails - They are troublesome and time consuming for the user - Internet access is allowed only for business purposes - They are useful for switched line access only (e.g. they offer no protection for - The protected network does not permit any download of executable programs access via X.25 or the Internet) from external sources - They offer no protection on the application or protocol level. - The network components keep log files on all Internet traffic. Dial-back systems in general are difficult to control and do not yet offer extensive security. 17.8.3 Access Control in Networks 17.8.3.1.1.3 Query/Reply Systems Data integrity and data security are ensured by a well thought-out access control based on a guideline defined in a security policy, for example, specifying what Query/reply systems consist of two partial systems, one installed on the user‘s user is granted access to what resources. Network access control is based on two machine the other installed on the host. When an access check is performed, the measures: authentication and authorisation. host system sends information to that of the user. This information is altered at the user side via an integrated algorithm and possibly completed with data. This 17.8.3.1 Authentication message is returned to the host system as the result of the operation. If the result is satisfactory, access to the host is granted. This process can be performed several Authentication checks and determines the identity of a user. After successful times in order to achieve even greater security. The hardware dongle systems authentication, administrators gain access to network components, hosts and integrated into some software products are also types of query/reply systems. network management facilities. Users gain access to the network and to the These systems offer quite a high degree of access security. However, depending on resources for which they were authorised. Authentication is especially important if the system type, there are certain disadvantages: public access facilities, such as access points, are implemented in the network. It is vital at such access points to distinguish users having access authority from those - The user may have to perform several operations who do not. - Depending on the type of system, they can be problematical to use The identification of a user is based on a unique code: - Apart from the PIN stored as an exchangeable token, a second PIN is needed for - which the user knows, for example a password or a PIN the host - which the user owns, for example a token or a smart card - Authentication systems can be loaned or stolen - which forms part of the user, for example a finger print - Export is problematical An efficient access control combines at least two of the above-identified criteria: Two Factor Authentication. These methods which are commonly in use 17.8.3.1.1.4 Software Tokens today in the EDP field, form the basis for a whole series of user identification and authorisation possibilities. The various procedures, together with their If access protection is integrated into the operating system or applications corresponding advantages and disadvantages, are described below. software, no external device is needed for authentication. This is the main advantage of software tokens. The disadvantages of this solution are: 17.8.3.1.1 Authentication Methods - They are dependent on the data end device and/or operation system - The software token can be copied or altered The most common procedures for user identification today are: - PIN, key and algorithm are all stored in the program - Independent passwords - Management is problematical. - Dial-back systems Audit trails and controls are arduous to perform and can point to the wrong - Authorisation by query/reply systems person. For this reason, there is no reliable way of controlling their circulation and - Software tokens use - Time-synchronised tokens . 17.8.3.1.1.5 Time-Synchronised Tokens 17.8.3.1.1.1 Re-usuable Passwords Time-synchronised characters or number strings which are constantly recalculated Use of the protection method most widespread today, i.e., passwords, has the using sophisticated algorithms, offer a considerably higher degree of security. The following disadvantages: hardware required can be accommodated on an area of the surface the size of a - Passwords can be manipulated credit card. - In practise, they are not changed at sufficiently regular intervals - They can be known to more than one user Advantages of this system: - They depend on different social components - Independent of the data end device - They can be decoded by special programs. - A card reader is not required Besides, their circulation and use are difficult to control and to track. - Login can be carried out in one step - No query/reply process 17.8.3.1.1.2 Dial-Back Systems - It is relatively secure - based on two independent factors (user‘s own PIN and token)- At any given time, only one person can have the token card; in case of Dial-back systems used with passwords offer greater protection. They have the theft, the card can be deactivated immediately. Thus, access can still be controlled advantage that access can be traced via the listing of telephone calls. An additional even when PIN and card fall into the wrong hands.

368 transtec = ttec in the Netherlands and Belgium be trulyeffective. Authorisation can be realised by implementing, for instance, for final access to a resource isfreed. C mechanismshouldbeconfirmedquestions. by anotheronebefore Onesecurity isalsotoshould beaccessible beconsidered to him.Redundancy in authorisation protection, onlythoseresources atask for absolutely auserto necessary perform userisalloweddefine how every to withthe work resources. To achieve effective user,assigned to every authorisationmechanisms limittheaccess to resources and Authorisation defineswhoisallowedrights Basedonthe to dowhat inanetwork. 17.8.3.2 Authorisation private keyisusedby theuserfor authentication. the password. The publickeyusedfor thisisstored on thetarget system, andthe authentication oftheusercanbedonewithencoding, inadditionto verifying through encoding, if, for example, auserlogsonto aUNIXcomputer. Here the protocolSSH isasecurity that ensures theauthentication andcommunication 17.8.3.1.2.6 Secure Shell(SSH) calculation. also allows for simplifiedmonitoring, statistical evaluations and, ifnecessary, canbeeasilyimplemented.Besides allthat, policy centralisationspecific security centrally, whichismucheasierfor theadministrator. Additionally, acompany- database canbefound centrallyDIUS, thiscanalsobeadministered usingRA contains userprofiles allauthorised isstored onthe central server. Since theuser communicate withacentralto authenticate server dial-inusers. Adatabase that protocolDIUS isaclient/server RA whichallows aremote to access server 17.8.3.1.2.5 Remote Authentication Dial-I authentication. TA formessage digestfunction passwords. TA to TA Although thenamemight causeone to thinkso, TA 17.8.3.1.2.4 TA passwords for userauthentication are thenadministered incentral database. toaccess forward authentication server data to aspecialauthentication server. The TA 17.8.3.1.2.3 Terminal Access C receives access authorisationinthethird step. password enters theencrypted andfinally the second, participant andtheexternal distant end, whichalsoprovides thekeyfor thecode, isaskedfor thepassword in step authentication process, logsoninthefirststep, where theparticipant the implemented by linksthat usethePPPtunnellingprotocol. This dealswithathree- The challengehandshakeprotocol ( 17.8.3.1.2.2 ChallengeHandshakeP network. The isclosediftheentriesare connection incorrect. the IDandpassword confirms thisandallows are correct, theserver access to the hasbeenmade,password theconnection after whichverifies it. toI theserver authenticationa two-step process, inwhichtheclient sendsanuncoded IDand by meansofanIDandpassword. I The P 17.8.3.1.2.1 P - IPaddresses are assigned to clients requesting access from acreditable centre following measures: security 17.8.3.1.2 Authentication P C A C CS isanauthentication protocol developed by C AP isasimpleauthentication protocol, that isusedto authenticate auser A CS. TA assword Authentication P C C A A CS+ isatotally newauthentication protocol that works witha CS+ C A CS+ usesthetransmission control protocol (T rotocols ontroller Access C t is part ofthePPPtunnellingprotocol.t ispart P entral control onauthorisationisvitalfor itto rotocol ( CHAP) isusedto authenticate systems andis rotocol (P n U C CHAP) A DIUS) (RA ser Service S+ also supports P CS+ alsosupports ontrol System (TA AP) C A isco, whichallows aremote CS+ isnotcomparable transtec = ttec in the the in ttec = transtec C AP andCHA A CP). CS) AP is f N etherlands and Belgium and etherlands sending device, which encrypts the data, over an unsecured network, for example for network, unsecured an overdata, the encrypts which device, sending of payload the encrypts is algorithm encryption used widely a of example Ancode. decryption the having users the i.e. parties, authorised by only decrypted be can effectively.done been data has The encryption the if reconstructed be cannot encryption. data. misuse and out read in, listen can attacker an that is transmission such in networks. unsecureddanger The P 17.8.4 P is raised andaccess specific host,to thehostisendedor analert refused. to beprotected. This technology comes into play whenanattack ismadeonthat I and/or thesessionisended. correspond to apredefined theNDISraises pattern analert orsuggestan attack, collision domain.As soonastheNDISrecognises ofpacketsthat apacketorseries systembased intrusion detection (NIDS)monitors allthepacketspassingthrough a There technologies. intrusion aredetection two complementary The network- conditions that have to bemetbefore data traffic isidentified asan attack. i.e. patterns ofattack. ofanattack typical ortype These signatures define certain I 17.8.3.3 I - Application filters allow onlypermitted applications - Address filters limittheaccess to resources tunnelling process inmore isdescribed detail. The GREtunnelpacketiscomprised P Encapsulation Routing 17.8.4.1.1.1.1 Generic 17.8.4.1.1.1 Tunnelling Standards andP take place onlayer 2andlayer 3. takeplaceagain. Authentication withinthetunnel. andencryption Tunnelling can of thetunneliswhere theIPheaderisadded, theendpoint iswhere itisremoved additional IPheader, inadditiononeormore specialheaderfields. The startpoint set upbetween theendpoints. Atunnelissetupby eachdata packetreceiving an I I protocol. This methodisemployed at thetransition from oneprotocol to another. same layer. The data of the oneprotocol are packedindata packetsofthesecond Tunnelling istheprocess inwhichtwo different protocols are encapsulated onthe 17.8.4.1.1 Tunnelling for mechanisms tunnelling users.use VPNs the of 3 or 2 layers on based usually is which network, logical closed, a is speaking, VPN Generally lines.dedicated through volumes,also data largefor and, network GSM or network telephone analogue the in access dial-up through either used transferred.be be can to VPNs is data when only up set are because advantages,VPNs cost yields also - used properly when - and flexibility moreprovides user, it the but by owned lines or lines leased with compared be can security,a VPN of viewpoint Fromthe network. public unsecured, an of use the with office head the to linked securely be to offices branch and of introduction The 17.8.4.1 P Virtual the n ahost-based system intrusion detection (HIDS),agents are posted inthehost attack signatures, systemsntrusion detection operate withtheuseofso-called nternet, between acentral VPN gateway andaremote linkis V PN client. Avirtual t serves for transporting data over for transporting t serves anunsecured, suchasthe publicnetwork, rotection of data transmission is necessary when data is transferredover is data when necessary is transmission data of rotection rotocol-independent tunnelscan besetupwithGRE I nternet, to the receiving device, which decrypts the data. the decrypts which device,receiving the to nternet, rotection ofData Transmission ntrusion D ncrypted data can be read out by third parties, but the contents the but parties, third by out read be can data Encrypted PN) rivate (V Networks reference model and is established for a specific group of group specific a for established is and referencemodel OSI etection etection V irtual irtual tunnel from the from tunnel a throughVPN sent is and packets IP P rivate N P s) enables mobile users mobile enables PNs) (V etworks rotection can be afforded only by data by affordedonly be can rotection rotocols traffic. IP network, or over the over or network, ISDN , astandard inwhicha Sec, which effectively which IPSec,

369

compendium of the tunnel and GRE header and the payload. The tunnel header contains the data in 64-bit blocks. The algorithm receives a 64-bit block that is not encrypted address information; the information on the encapsulated tunnel protocol is in the and then outputs a 64-bit block that is encrypted. DES is a symmetric algorithm. GRE header and the encoding algorithms and payload cover the protocol header Encryption and decryption uses the same algorithm and key. The key length is 56 of the encapsulated tunnel protocol and the user data. bits. The key is usually referred to as having 64 bits, but every eighth bit is used for a parity check and is ignored. These eight parity bits are the lowest-level bits in the 17.8.4.1.1.1.2 Point to Point Protocol (PPP) individual bytes for the key. Every number that is 56 bits in length can be used as a key, and the key can be changed at any time. The DES standard was superseded The point to point protocol (PPP) was created to encapsulate datagrams over in 2001 by the AES standard, since the so-called DES crackers compromised the serial connections and supports transmission of all common LAN protocols. integrity of the key. The PPP protocol overrides the limits of interoperability, which is caused by the encapsulation technology used by bridges and routers during transmission 17.8.4.1.2.2 Triple Data Encryption Standard (3DES) over WANs. The PPP protocol enables data transmission over synchronous and asynchronous dial-up and dedicated lines. It is thus able to work independently The triple DES key refers to multiple encoding based on the DES. The symmetric from the respective physical interace. The only prerequisite that is necessary for encryption algorithm 3DES uses two keys and three DES cycles. 3DES thus works using the PPP protocol is a completely transparent, full-duplex-capable data circuit. with a 168-bit key length, or 56 bits for a DES cycle.

17.8.4.1.1.1.3 Point to Point Tunnelling Protocol (PPTP) 17.8.4.1.2.3 Secure Socket Layer (SSL)

The leading point to point tunnelling protocol, developed by Microsoft, was SSL was specially developed to encrypt information in the Internet and is based on recommended in 1996 by the IETF as the standard protocol for Internet tunnelling. TCP/IP. SSL encodes with public keys which are usually acknowledged by a third PPTP is an expansion of PPP. PPTP encapsulates PPP packets in IP packets, so that party. High security is guaranteed, since the key for decoding must be individually protocols such as IP, IPX and NetBEUI can be transmitted over the Internet. PAP and set and is only stored by the user and not transmitted over the Internet. The CHAP are used to control access. The data encryption standard (DES), with keys developers of SSL have designed the protocol to have two levels. One level is that have a depth between 56 (DES) and 168 Bit (3DES), is used as the encoding for encrypting the data. It permits various algorithms, including DES and 3DES, algorithm. With PPTP, the user is responsible for controlling the end of the tunnel. and requires both communication partners to have the same secret key that is generated for every connection. Data authenticity is also verified by a checksum 17.8.4.1.1.1.4 Layer 2 Forwarding (L2F) test. The private keys are exchanged on the second level. Participants for the communication are authenticated, arrange an encryption algorithm and then send Layer 2 Forwarding (L2F), developed by Cisco Systems, is specially used to connect each other the encoded session key. individual computers. In conjunction with PPTP, it forms the foundation for the layer 2 transport protocol (L2TP), which is a further development of both systems. 17.8.4.1.2.4 Advanced Encryption Standard (AES) L2F supports various protocols and several independent, parallel tunnels. User identification is somewhat weaker than that for PPTP and extra data encoding is The advanced encryption standard was developed in 2001 to deal with security not provided. problems that arose from the DES standard (“DES crackers”). The AES is based on the Rijndael key and thus recognises three key sizes with 128, 192, or 256 bits. 17.8.4.1.1.1.5 Layer 2 Tunnelling Protocol (L2TP) The AES is so secure, that it can create 10 to the 21st power more 128-bit keys than the DES 56-bit key. The layer 2 tunnelling protocol (L2TP) is only different in a few respects from PPTP. On the one hand, L2TP supports several tunnels, just like L2F. On the other hand, 17.8.4.1.3 VPN Configurations the user does not control the end of the tunnel as in PPTP, but is done by ISP. The following section deals with the configuration and setup of VPNs for various 17.8.4.1.1.1.6 IP Security Protocol (IPSec) applications.

The IP security protocol (IPSec) was specially developed to connect two local 17.8.4.1.3.1 End-to-End VPNs networks. In doing so, the IPSec protects the data packets of the IP protocol from possible modification or copying. IPSec influences neither the communication End-to-end VPNs provide a direct connection between several workstations. This protocol, nor the application program, so that routing is not interfered with. type of VPN can be used, for example, to safely connect customers with an online Authentication processes, that have been created with IPSec, can differentiate shop or to help employees at different locations to work on the same project. between data from authorised and unauthorised communication partners. The When using end-to-end VPNs, you must remember to install the corresponding authorisation processes are based on MD5 hash algorithms with 128 bits and VPN protocol on the connected computers, since the workstations are directly the encoding on the DES algorithm with 56 or 168 bits. All data traffic can be connected to each other and not via a VPN server. L2F, L2TP and IPSec are protected with IPSec and even Ipv6 can be transmitted via IPSec. particularly suitable protocols for setting up end-to-end VPNs. IPSec is, however, best suited for applications that require the highest level of security. 17.8.4.1.1.1.7 Layer 2 Security 17.8.4.1.3.2 End-to-Site VPNs The layer 2 security protocol (L2Sec) is supposed to eliminate specific weak areas that are exhibited by IPSec in remote access solutions. With L2Sec, all data packets End-to-site VPNs, or Remote-Access VPNs, are mainly used to connect external are packed into a tunnel and then saved as a whole. employees to a central company network. The greatest advantage of such a network is that the employees can dial into the network using any POP from 17.8.4.1.2 Encryption Standards the company’s service provider. End-to-site VPNs can help to reduce the costs of expensive long-distance connections, since you do not need to provide a large 17.8.4.1.2.1 Data Encryption Standard (DES) modem pool for employees to dial in.

Data Encryption Standard (DES) refers to a block encryption, which encrypts the

370 transtec = ttec in the Netherlands and Belgium 17.8.4.2.2 Remote Node high travel costs incurred for visits by anexpert. inexpensive problems, meansofsolvingsoftware andfrequently dispenseswiththe of asinglecomputer oraLANvia WA N link. This provides simpleand avery number ofgraphics. Onemajorapplication ofRemote C the applications useddonotinvolve frequent complete screen refreshes oralarge remote control technology relatively permits highprocessing speedsaslong dataBecause transmissions are limited to keyboard andscreen display information, system, remote operation isexclusive orcanbeparallel. D computer canbeastand-alonedevice oraworkstation withinalocal network. controlled by asecond computer viaa WA N connection. The remotely controlled Remote C 17.8.4.2.1 Remote C system. other any to apply also generally system operatingbelow. consideredthis forbe made will statementsThe cases these only a system,Windowsunder operating run systemsaccessing of majority great the However,end. each atdifferent systemssince with feasibleare applications these categories: main threefollowing the of one intoTypically, classified are applications accessremote compression. data of means by transmission data acceleratesthe protocolthis connections, available,e.g., serial is bandwidth small a only where P P and elsewhere.protocolsSLIP older The of setup synchronously.and operandi modus The as protocolsT such protocolforlines. line WA serial overspecial a connections is N network. the via or systems the between data a prerequisiteis second The PDN). (V networks dial-up private virtual the through connections The used. be also above,can sections the in FrameRelay, or described alreadyX.25 GSM, as such connection serial or controllers them. to connected modems with used be also systems.However,can both servers communication on interface serial a to connected modems via is connection a establishing of way located.are easiest applications The or data the which on system the and access a prerequisiteis first The below.detail in described are which factors of range a requiresaccessSuccessful LAN. a via computers these with connected directly not is which system a fromcomputers on data or applications to access means access Remote 17.8.4.2 Remote Access company remote access. network, All theprotocols referred to here canalsobeusedfor secure to a connection secure to acompany connection (remote network access). orpublicutilities.networks All theprotocols referred to here canalsobeusedfor are withthismethod. connected This configuration issuitable companyto connect Site-to-site VPNs are theclassic VPN variant. Several LANsfrom variouslocations 17.8.4.1.3.3 Site-to-Site VPNs which itisdialling. C For aremote nodesolution, theremote PCbecomes into anodeinthenetwork rotocol (LBX) has been developed fordeveloped W been Windows-basedsystems.has Forcases rotocol(LBX) X LowBandwidth nowadays. importance so-called Thelittle of rotocol)are epending onwhethertheremoteoramulti-user computer isasingle-user I nternet became ever more important as an economical transport medium for medium transport economical an as important more ever became nternet ontrol isusedto operations describe inwhichacomputer isexternally Remote CP/IP ontrol terminal adapters are used. are adapters terminal ISDN ommunication takesplace, asinaLAN,viathenetwork I f is selected as a transmission service, internal service, transmission a as selected is ISDN ,

network connection network IP C PPTP ontrol,Remote X or X communications protocolcommunications . allows setting up secure tunnels for so-called so-called for tunnels secure up setting allows PPTP DEC net to be transmitted asynchronously or asynchronouslytransmitted be to net ( CSLIP

Node between the computer seeking seeking computer the between is most important here. important most is PPP C and ompressed Serial Line Serial ompressed is described in further detail further in described is PPP Data Transfer I n principle other types of types other principle n which is used to transport transport to used is which ontrol isthemaintenance transtec = ttec in the the in ttec = transtec I t enables LAN enables t

. All of All . I nternet ISDN PPP

N etherlands and Belgium and etherlands The modemwillcontinue to beasignificant data transmission medium for aslong M 17.9.1 Analogue M 17.9 Hardware ina WAN system plusa Web browser via T e.g.tasks runninginparallel viaaPPPconnection, atelnet window for aUnix IP Authentication P using P controlextracting characters, orpassword monitoring thelinequality validation parameters,negotiating theconnection compression, determining theblocksize, available LANprotocols. data varioustasksduring transfer, PPP performs suchas PPP isnotanotherLANprotocol, protocol butisessentially acarrier for the PPP istherefore data together withtheLANprotocol. to transport C which ismostcommonly used, isthe line,a serial aspecialcommunication protocol orcarrier isrequired. The protocol, As LANprotocols suchas T comparatively slow line, itissensibleiftheyare alsoavailable locally. that theapplication programmes donotalways have to beloadedover the With Remote Node, allapplications runontheaccessing computer. To ensure either assigned whendialling-inorfixed inadvance. PC therefore hasitsown address network nodeinthenetwork. like a direct This is protocol operated protocol by meansofacarrier viathedial-upline. The remote used. normally are modems asynchronous consequently and modems synchronousfor required nevertheless protocolsareappropriate the with connections synchronous lines. corresponding Theclocking separate via achieved Synchronisationis stream. sequential continuous a in sent are bits data the of all transmissions synchronous bits. stop two or one and bit start supplementary a assigned is character unit. data basic the as characters uses transmissionAsynchronous rate.transmission and synchronous) or (asynchronous mode transmission their are market the on currently modems of variety wide the distinguish that factors major two The 17.9.1.1 Transmission M commands are prefixed withtheletters AT. and controlling modems. I The Hayes-compatible command sethasbecome thestandard tool for configuring have effectively driven separate, individual faxcards from themarket. modemscanalsotransmitBecause andreceive facsimile(fax)transmissions, they personal computers are equippedwiththecorresponding connection. (PCMCIA). interface PC-card These popular, are very portable allmodern asvirtually interface. serial the RS-232 Yet another, more recent alternative isamodemwith devices, inthePC,ortheycanbeexternal whichareto beinserted via connected according ofmodem;modemscanbeinstalled asslotcards to thetype designed reconverting thembackinto digital format. The installation arrangements vary telephone lines. The receiving modemdemodulates theseanaloguesignals, converts theminto analoguesignals suitablefor transmission through the The modemreceives computer data from thecomputer indigital format and answering machinefunction. andanautomaticcommunications PC-afaxfunction ontheconnected software such aserror data correction, compression and-together withappropriate dem M the ISDNnetwork. terminal viatheanaloguetelephone rather network thanhaving access direct to as alarge worldwide are numberoftheinstalledconnections sent to theuser X orAppleTalk canalsobeusedsimultaneously. I odems are usedfor data transmission intheanaloguepublictelephone network. odem isaword coined of from theoriginal functions odulation. Today‘s modemshave anumberofadditionalfunctions

AP ( assword Authentication P rotocol). With PPP

odems odes t isalsoreferred to astheAT command set, asthe CP/IP , IP CP/IP , different protocols network suchas T X orDEC . P rotocol) andCHAP( oint-to-P net cannotbetransferred over directly oint-P t isalsopossibleto have several rotocol (PPP) mo Challenge Handshake dulation and . The taskof ach transmitted Each onsequently, CP/IP I n , 371

compendium The unit used to measure a device‘‘s transmission rate is the number of bits per and network load distribution. Routing becomes necessary when communication second (bit/s). The baud rate, as a unit, indicates the signalling rate over the between stations in different subnetworks needs to be established. transmission route. However, as more recent transmission methods no longer merely use frequency but also amplitude and phase to convey information, several Routers are not protocol-transparent, but must be able to recognise all the bits can be transferred per baud. protocols used, since they translate information blocks according to protocol Modems can never be fast enough for transmitting large data volumes. The high specifications. Thus, the protocols used must be either routable or be procurement investments for faster modems will soon pay for themselves due to correspondingly translatable, i.e., re-packageable into other protocols. Since not all the low operating costs. protocols can be routed, most routers are also able to bridge packets. In comparison with switches, routers ensure improved isolation of the data traffic, Even the V.34+ transmission standard, which specifies a gross transmission rate of since for example they do not forward broadcasts as standard. However, routers 33,600 bit/s, has not yet reached the maximum capacities of traditional telephone generally slow down the data transfer. Still, in branched network connections, lines although experts predicted that a transfer rate of 28,800 bit/s would be the especially in WANs, routers actually transfer data more efficiently.O n the maximum. The speed category 56,000 bit/s is standardised as V.90. Using V.90 other hand, routers are usually more expensive than switches. Therefore their allows a download transmission rate of up to 56 kbit/s. A prerequisite for this is a implementation should be carefully analysed for each specific case. so-called 56k host. The higher transfer rates are achieved by dispensing with the analogue conversion of the signal between the 56k host and the digital switching The logical addresses in a network can be evaluated by routers and the optimal centre. Only the user‘‘s connection continues to be analogue. path (route) from sender to receiver can be determined with the aid of internal Older, manufacturer-specific protocols such as PEP, HST, V.32terbo and V.Fast are routing tables. Routers adapt the packet length of the data to the maximum no longer of any relevance, since the V.90 standard surpasses all of them with its possible within a network segment, i.e., they change the packet length for example transfer rate. at the transition from Ethernet to X.25. Routers not only modify the packet length, they also adjust transmission speed at the transition point between LAN and WAN. 17.9.2 ISDN Adapters To this end they need a suitably large buffer, which in most models can be flexibly configured. Communication between existing applications is generally adapted to serial analogue modems which recognise the Hayes AT command set. To use ISDN, in Many different types of routers are available.Local routers are primarily used such cases, the application of an ISDN terminal adapter with a serial interface for security considerations, e.g., if communication in a LAN is to be restricted to is recommended. As the ISDN terminal adapter is an external device with a specific nodes. Alternatively, they are used if the data load on a network is to serial port in accordance with RS232, it can provide a direct replacement for an be reduced, that is to say, when collision domains and thereby the number and analogue modem. Similarly to the analogue modems, control is performed using distribution area of broadcasts are to be minimized. Local routers are available for the extended AT command set. For this reason, ISDN terminal adapters are suitable, all Ethernet speeds. in principle, wherever data is to be transferred from the serial interface via ISDN. Examples of usage include workstations without an internal ISDN controller, as Due to the large number of ISDN lines available and ISDN‘s attractive tariffs, well as remote bridges and routers with a serial interface. ISDN terminal adapters ISDN routers are often used as remote access routers. However DSL and, given are also suitable for use with servers to implement high- commensurate data volumes, transmission technologies for up to 2 Mbit/s are also performance remote access applications via ISDN. Note that not every serial used for remote access. To ensure redundancy and thereby high availability, ISDN interface is appropriate for the high transmission rate of 64,000 bit/s (up to connections are provided as backup for remote access routers of high transmission 256,000 bit/s or more with channel bundling/data compression) used in ISDN. rates. Refer to the technical specifications for the particular interface for more detailed The router types described so far are all based on a combination of hardware information. In any event a serial interface with incorporated data buffer (UART and software, but there are also pure software solutions running on servers, 16550-compatible interfaces) is of advantage. workstations or PCs. What solution is most suitable for a specific application must be evaluated for each individual case. It should be added that optionally most ISDN terminal adapters can also be The basic components and the technical characteristics of routers are discussed operated through a CAPI interface in addition to the AT command set. More below. information on CAPI can be found in the CAPI section. 17.9.3.1 LAN Interfaces Common telephones, answering machines, group 3 facsimile devices and modems are analogue end devices and cannot be operated directly on the ISDN connection. Most routers are provided with one or more LAN interfaces, according to topology In these cases, an ISDN terminal adapter with an a/b interface is the solution. for Token-Ring, Ethernet, 100 BASE-T Fast Ethernet, FDDI or ATM. For connecting It allows the use of available analogue end devices even with ISDN. Small ISDN corresponding media either multipurpose ports (e.g. Ethernet AUI, BNC, RJ-45) are PBX systems offering up to 16 analogue PBX lines are, in principle, equivalent available or the connection is implemented as a slide-in unit which can therefore be to a 16-port terminal adapter. However, the analogue lines limit the quality and customised for the requirements. transmission speed of this type of device. 17.9.3.2 WAN Interfaces ISDN controllers represent the most inexpensive and simplest option for entering the world of ISDN data communication. They can be obtained for Today, a number of companies offer WAN lines supporting different transmission systems and hardware platforms. Transmissions through BRI as well as PRI are speeds. The tariffs as well as the interfaces offered vary correspondingly. possible, depending on the card. However, the most common form is PCI insert For smaller networks (e.g. workgroups), connections with transmission speeds of cards for PCs. 64 Kbit/s are advisable. There are, of course, applications that can manage with lower transmission rates. If higher transmission rates are required, the best solution 17.9.3 Routers is the use of E-1 connections (in principle ISDN primary multiplex connections), which are common in Europe and offer a transmission speed of 2,048 Kbit/s, or Routers, in contrast to bridges, also link networks of different topologies in OSI connection via DSL. layer 3. They are the cornerstones of well structured LANs and WANs. Their ability Routers are equipped with one or more WAN ports, which are either built-in, or to route different protocols and network types provides optimal traffic control in the case of modular versions can be upgraded with corresponding interface

372 transtec = ttec in the Netherlands and Belgium implemented onrouters. made onpacketfilters are relatively smallandtherefore packetfiltersoften are senderandreceiverchecking address andtheIPservice. The hardware demands P 17.9.4.1.1 P 17.9.4.1 Firewall Technologies simplified by concentration ofaccess onasingle component. management, aswell asmonitoringSecurity andcontrol are functions, significantly individually configured dependingonuser-requirements andsecurity. for services A firewall generally consists ofmultiplehardware andsoftware components contravention. is interrupted and, is raised incaseofserious dependingonconfiguration, analert firewalls allow passageorblockdata. I passage onlyto approved services. The firewallcriteria rules definethe bywhich the I of variousmechanisms. They assumethecentral control ofcompany access to andarenetworks intended to inacorporate increase security by network means Firewalls are computers whichoperate gateways asdedicated security between 17.9.4 Firewalls expanded ifneeded, orcome withthecomplete package. software the manufacturers. Routers are eithersuppliedwithabasicversion that canbe executed. The supplieddiffers range ofsoftware according to thestrategies of When bootingtherouter, isthenloadedinto thesoftware theRA isgenerallyRouter stored software inFlash PR 17.9.3.4 Software bridged, provided therouter hasthisfunction. addition to IPand DEC inadditionto IP may besupported and theequipment, IP most routers alsooffer multipleprotocol capabilities. D by routerssupported isIP Routers route oneormore protocols. network The protocol network mostoften 17.9.3.3 P ISDN connection. X.21; for theasynchronous mode, theRS232interface. isusedforThe S0interface modules. Standard physical for synchronous interfaces modeare RS449, V.35 and acket filters monitor at IPlevel whetherornotapackethasaccess authorisation, net, AppleTalk, OSI,XNS, VINES andApollo D nternet by authenticating andallowing usersonthebasisoftheiridentity rotocol Support rotocol Support acket Filter X. All non-routable protocols suchasLAT andNetBIOSwillbe X, DEC . D ue to thewidespread useofheterogeneous networks, net, AppleTalk, OSI,XNS, VINES andApollo D . D epending ontherouter andtheequipment, f sucharuleiscontravened, thedata flow OMs, easyupgrading. whichpermit omain may be supported in omain may besupported epending ontherouter transtec = ttec in the the in ttec = transtec M and omain N etherlands and Belgium and etherlands The application gateway receives thedata I packetsat therespective ports. physically identification user. prior andauthentication andexpect ofevery are implemented across multiplelevels. They candecouple logically networks and alternative to afirewall. I Application gateways represent themostsecure, butalsothemostcomplex 17.9.4.1.4 Application Gateways internal IPaddresses canberead outfrom outside. no oftheseparation network, ofinternal andexternal virtue authorisation. By hastoto register setupaconnection withthehostandprove ithasaccess filter, butoperates at ahigherlevel oftheprotocol A computer stack. wanting communication takesplace. The circuit relay concept iscomparable withapacket andaninternal routerwith asubnetandanexternal withahostthrough whichall C 17.9.4.1.3 C are to superior pure packetfilter firewalls. security-relevant applications, firewalls incorporating statefultechnology inspection determining thecorrelation between data packetsbelonging together. Therefore in data packetsare madeonthebasisofcomparing multipledata packetsandby level andare stored indynamicstate tables. The decisionsonforwarding the the data packetsare analysed, transmission, during onthecommunication statusanalysed isincludedinthedecision.I andtheconnection Stateful isapacketfiltering technology inwhichthedata packetsare inspection user isnever established. mediator between destination oneithersidesothat computer aconnection and I onadestination process computer.communicating withtheserver oftheservice to theother. From theviewpoint oftheaccessing user, itappearsasifhewere andtransfersthe application server the packetfrom onesideofthenetwork isactivated on (aproxy software server) isto bepossibleviatheport, a service 17.9.4.1.2 Stateful I n point of fact, however,n point offact, as theusercommunicates acting withtheproxy server ircuit level gateways associate packets to existing T ircuit Level Gateways nspection nspection n application gateways, mechanisms adequate security CP connections. They operate n thistechnology f only 373

compendium 18. LAN Core Solutions

18.1 Introduction traffic demand due to new locations and new users, yet without imposing any constraints on future growth. Such demand often comes entirely unexpectedly. Nowadays fast communication and common access to information, ideas and When, for example, a large number of contractors work together on a large resources is becoming increasingly important. For this reason data networks are project, this involves an increase in data traffic. These short-term peak loads vital for business success. should not impact the long-term network strategy. Furthermore, the operator has to be in a position to set up new technologies cost-effectively. And, at the same This economic scenario requires modern companies to be more competitive and time, the focus should be on secure connections and more cost control, despite to increase their productivity. Optimisation of the data flow within the company constant expansion or redesign of virtual LANs. as a whole has become an important constituent of this process. The provision of What are the crucial economic factors in developing or selecting core bandwidth-intensive content and sophisticated multimedia applications calls for technologies? Here consideration has to be paid to the individual requirements of efficient use, signifying, in turn, the need for a stable network infrastructure. each company. These new requirements mean that networks are designed in such a way that they support the current business operations, but are also geared to future 18.2.2 Cost Control growth. The increase in switched networks, substantially reducing network complexity, is an important development for companies of every size. Routing The requirements are: achieving more and more and securing growth with tight technologies formerly playing a predominant role are taking a back seat. With resources. That is why low purchasing costs are a vital factor. In the present-day high-performance LAN traffic, switched networks provide a cost-effective basis for business environment it is equally important for the network to be geared to the providing applications of the following generation, including video streaming and future for the long term. The overall operating expenses should form a significant sophisticated Web services. part of the equation. If it is possible to achieve savings here and at the same The currently available Layer 3 switches perform all the operations of a time ensure fail-safety, this has a significant impact on the return on investment. conventional router. They can be used at any point within a network core Companies then secure better control over their expenditure. or backbone and operate in the network layer of the OSI (Open Systems Interconnection) reference model, ensuring significantly higher data transfer. 18.2.3 Scalability and Flexibility The new switch generation has been developed specifically for operation in higher OSI levels. It boasts even more flexible functions and hence copes with a The scalability of networks is an important factor in the overall resources. Where wide spectrum of data traffic modes. In addition, it enables enhanced network extra functionality is needed, special requirements and funds should be carefully management. The latest switches are even more intelligent and achieve a higher balanced, thus optimizing future-proofing of a network. In many companies the throughput. They offer substantial advantages for all network types, for example demands on capacity are dynamic and unpredictable. That is why flexibility figures user authentication and Quality of Service (QoS). Layer 4 switches, which identify large in the configuration of the network. The set-up of a scalable network gives and forward data traffic on the basis of multilayer packet information, permit data companies maximum security for existing investments. traffic administration at network level and not just in the central core. The advantage: faster replies as well as maximization of server use and availability. 18.2.4 Security These enable you to identify and establish priorities for critical data traffic based on application. Switched networks indeed have their advantages, but they have involved a heightened security risk at the very point where the network is not protected. In the last ten years companies have been constantly harnessing new potential End-to-end security has today become a critical task in network administration, for building and for the administration of corporate networks. This development particularly when the phenomena accompanying e-business, the use of extranets is forging ahead to support critical business requirements. Switched networks are and the increase of data traffic in private networks are included. Today networks important to ensure optimum data access and services. At the same time they have to protect users, systems and intellectual capital more than even before. This are guarantors for high-grade security, mobility, provision of information and, calls, above all, for extended authentication and control. what is so important in the present-day business world: teamwork. This chapter is intended to give you an understanding of the technologies currently available. 18.2.5 Availability

18.2 New Strategies for Networks The workstation today looks entirely different from a few years ago. It is no longer simply a nuisance if the network becomes slower or even fails, it can seriously Nowadays a company relies on the network living up to the demands posed jeopardise critical business operations. Most companies nowadays employ by daily business life: canvassing customers, transacting business and gaining a hardly any IT staff trained in system administration. Network failure therefore competitive edge. How can a network ever keep pace with the demands of the has catastrophic consequences. Resilient network infastructure is indispensable daily growth in business - especially without exceeding the financial limits? for problem-free communication and uninterrupted operation. The network also has to be capable of self-regenerative operation, it has to display high- These are the challenges confronting many network managers today. At the same performance and support a multitude of applications and high data traffic. High- time there is a call for the business requirements to be met within the framework grade availability and fault tolerance depend on the configuration and functions of the budget available. This can only be done if the planned network is of the devices distributed over the entire network. To ensure access to critical expandable without going to great expense. Future network expansions should be network applications, chassis-based switches and stackable configurations are of possible without major restructuring, which is costly. To achieve this, the selection advantage. of the network core is decisive. 18.3 Current Solutions

18.2.1 Structure of a Network Core The network capacity in very large companies is as a rule predictable or static. The Changes in corporate structure, whether they take the form of a merger, takeover network core is affected only by small or scheduled modifications. These generally or simply growth, put considerable demands on the network backbone. In an require the scalability, availability and performance of a centralised core backbone. ideal case, the potential afforded by the budget should meet the increased data Large networks demand extended multilayer switching functions in order to

374 transtec = ttec in the Netherlands and Belgium the network, without the physical limitations of a centralized core device. Coredevice.centralizedcore a of limitationsphysical the without network, the growswith that fabric switching distributed managed,centrally a as behave switches Gigabit interconnectedemployed, multiple is technology XRN When investments. network safeguardyou and costs up-front only! the reducesThisdemand on expansion signifies: technology XRN routers.of protocolsand scalability connections, redundantThe offering availability,high of backbones core high-performance of design the for managers network to scalability practical deliver switches XRN-enabled of functionality featuresand fabric. performance distributed The a of concept this on based solution new a is 3Comfrom Networking) Resilient (eXpandable technology XRN reduced.substantially is failure network complete of danger functions. the Thus redundancy complex and expensive with dispenses configurationcore flexible A failure.of point single a have not does core the that ensure can one switches, several between them distributes and platformcore central,high-end single, a fromcore the of performance and intelligence,functionality the withdraws one If devices. various the of way by managed is routing and switching all distribution, of degree high a with network a In steps! in expand they meaning needed, are ports further or performance in increases whenever devices add can providers network wayhierarchy. this of instead aggregation In of principle the on based is It point.controlcentral a on supported longer no is network distributed A networks.core distributed torelates approach This access.remote secure,cost-effectiveprovide and teleworkersprogrammesfor support they time same the Atbranches. of management easier securethereby Companies wanted.data the to accessimmediate user every offers and network entire the overdegreeincreasing an to spreadsinformation the Hence performance. end-to-end and video or voicefor prioritization service ensure now network the of sector edges the in devices intelligent further and points accesswireless switches, Scalable edges. network of direction the in moving now are switchcore the with only associated formerly functions Important bandwidth. output and scalability for need the with market the hitting are devices of generations New withDistributed Core18.4 Networks below.explained is done be can this growth.How future and operations business present the of support between equilibrium realistic a attain to as so important is architecture network dynamic a companies Fortarget.growingreal the be should network, entireavailability, the covering end-to-end with strategy a companies Formost requirementsbest. the meets that type network the select to important is It solution. right the is chassis available,a requirementsis the matching ports of number the having device network a and accuracy with predicted be can capacity in increaseanticipated the If switch.configured similarly second, a need will you further,expanded is it when switch the of capabilities the outstrips network the higher.are costs If,upfronthowever, the full, to used not is capacity the When selected. is chassis appropriate most the that so advance in planned thoroughly be should capacity network chassis.in increasethe Theto modules extra add just you added, are modules.users When interface for slots several with fabric routing a comprises example.implementationfollowing chassis-based the Theat look us Let overloaded.alreadytomorrow is and use insufficient to put today is backbone conventionalrequirements, network a capacity unpredictable with size medium of companyexpanding rapidly a In solution. effective most the certainly protocols(e.g.IP routingfor support demand networks capacity.storage and When density port functionality,processing necessary the companies offer switchescore video.high-end Centraland communicationvoice as such traffic data real-time prioritize to and applications sophisticated support hence provide for top performance. Every switch in the fabric constitutes a fast a constitutes fabric the in switch Every performance. top forprovide hence and fabric distributed the within are authentication and QoS as such functions

X and Appletalk), this is this Appletalk), and X transtec = ttec in the Netherlands and Belgium and Netherlands the in ttec = transtec

RR is a sophisticated routing implementation. routing sophisticatedTwo in a switchesinterconnected is DRR (DRR) 18.4.1.2 Distributed ResilientRouting resilience.and performance enhanced the from benefit immediately and fabric distributed XRN the to connected directly are standard this aggregation. link 802.3ad IEEE standard-based supports DLA ensured. is capacity available of utilization Optimum switches.core both to paths with configured be present.at units two to fabric distributed XRN an in extending ports via aggregation link implementing of possibility the offers DLA 18.4.1.1 Distributed LinkAggregation (DLA) 18.4.1 XRNComponents overheads.administration reducing easier,resourcesand much IT freeingis network the entity,across administration single a compose switchesinterconnected the Since backbone. core the for switches.interconnectedredundancy router the ensures Thisbetween forwarding high-performance permits also technology routingresilient distributed The availability.and performance superior provides this aggregation), link (distributed fabric distributed entire the in aggregation link supports technology XRN As resilience.and unprecedentedperformance are:consequences The load. routingentire the bear tohave not do traffic.processors data individual The the forwardsphysically fabric the in switch every as network, the of expansion the with line in increased entity.is switching 4 Layerperformance The and 3 Layer users and key applications require redundant connections to the core switch and switchcore the to connections redundantrequire applications key and users Special expansion. business enabling solution, network redundant and scalable highly high-performance, resources.a management offer simultaneously They and administration network in outlay minimum demand usually networks Basic 18.5.1 BasicNetwork requirements. dynamic confrontedwith managers network to design network forpotential new entirely up opens fabric distributed with network a of implementation The upaDistributed Fabric18.5 Setting inaNetwork switches.interconnected the of one in outage an of event the in monitoring and management device continuous ensuring units, both across shared is address IP management the addition, overheads.In administration and complexity minimizing fabric, distributed the across performed all areconfiguration QoS and filtering multicast modification, parametertree upgrades,spanning configuration, software VLAN as such tasks, Management SNMP using address IP single a via entity logical single a as managed be to fabric distributed connected, are devices edge which of Regardless fabric. distributed XRN the acrossinformationcontrol and management distributing forresponsible is technology. XRN It of systemcontrol the is DDM 18.4.1.3 Distributed D benefits. performance significant in results also this devices, connected for forwarding local 3 Layerprovides fabric the in switch every reconfiguration.Since end-station requirementfor any eliminating and outage over, takes network automatically switchaverting other failure,the switch of event the requirements. In dynamic with faced managers network for up opens design network forpotential capacity.new Entirelybandwidth of optimisation and performance routine of utilisation Maximum fabric. distributed the in switches both across capability routing the of distribution intelligentfor entity.routing active single a as behave fabric distributed XRN an , TELNET or Web, management. TELNET

evice (DDM) Management

LA permits layer switches to switcheslayer permits DLA M allows the entire XRN entire the allows DDM D evices supporting supporting evices RR provides DRR 375

compendium to business-critical servers. Low costs are a further decisive factor. For extended networks, a Layer 3 Gigabit Ethernet network on four levels, A typical example of appropriate architecture would be a Layer 2 infrastructure comprising workgroup aggregation, core connectivity and server aggregation, with a network architecture on two levels - Layer 2 - Layer 2 (i.e. workgroup is the right solution. This ensures the acceptance of legacy protocols and the switch to core switch), scalable and capable of expansion to form a full IP routing transition to a seamless IP infrastructure. network without the otherwise usual complexity. For basic networks the implementation of a distributed fabric offers higher Advantages: security for network availability. In case of a further business expansion and - Complete redundancy of edge, workgroup, core and at the server avoids a increasing user numbers or subdivision of users into different departments, with single point of failure core switches Layer 3 switching software is easily activated and thus ensures - Stacking resilience minimal interruption of operations. Additional network components due to - Trunking at stack level (dual connection to the workgroups) increasing office area - whether permanent or temporary - can be administered - DLA (distributed link aggregation) supplies XRN resilience more cost effectively. Furthermore, authentication of the users before they access - Redundant switching fabrics offer complete redundancy and enable rapid network resources is possible. switching Advantages: - Client availability through VRRP support in the core - Delivers the right performance level for the given applications at low cost - Dual connection of the servers to the server aggregation switches ensures 100% - Supports network telephony where needed availability of the server farm - Low demands on resources for network administration and maintenance - Prioritisation of the desktop switch applications makes for end-to-end - High network availability from the edge all the way to the core perfomance - High resilience through the deployment of appropriate workgroups and core - Network telephony is realisable immediately switches - Efficient utilisation of WAN bandwidth, the desktop and workgroup switches - Dual home trunking function provides for comprehensive networking automatically rerouting the web data traffic to a web cache and thus cutting overall WAN costs. 18.5.2 Standard Network 18.6 A New Direction for Corporate Networks Standard networks are usually based on the IP protocol. They often have to be converted to Gigabit Ethernet in the core or at the aggregation points while The further development of switched networks is running at full speed since safeguarding existing investments. The more intensively Internet is used, the there is a rapid increase in bandwidth demands. Therefore the need for expanded more important are redundancy/resilience as standard and an intuitive network network functions must be met. Network connections continue to be wired or administration. Standard networks often extend to several buildings in proximity wireless. Flexibility and mobility must be ensured without detracting from security as well as to remote sites. A common network topology is a network architecture or user-friendliness, enabling the user to establish connection to a network from on three levels - Layer 2 - Layer 3 - Layer 2, comprising workgroup aggregation, remote sites. core connectivity and server aggregation. The wireless „roaming function“ can also be implemented at the company A Layer 3 Gigabit Ethernet solution on three levels consists of workgroup premises. Fast and secure access to business applications and services can be aggregation, core connectivity and server aggregation. gained from any location. To cut overall operating expenses, 3Com‘s solutions support a new direction Advantages: in corporate networks. Earlier, network intelligence was available only to core - Economical performance in Wirespeed from the edge all the way to the core devices, today any switch, wireless access point and further endpoints of the next with 2:1 blocking generation can tie in, bringing functions like traffic prioritisation, authentication - Supports network telephony where needed and encryption closer to users. - Low demands on resources for network administration and maintenance „A pay as you grow“ cost structure is the result of an intelligence distributed - High network availability from the edge all the way to the core across the network components. It represents improved cost control and - High resilience through the deployment of appropriate workgroups and core administration as well as targeted provision of new services and functions. switches The enhanced interoperability between the various network devices permits more - It is easy to assign priorities to applications and to block them in the network effective use of existing network investments and the addition of voice, data or - Dual home trunking function provides for comprehensive networking. wireless elements where needed. Diverse applications, including Voice over IP (VoIP), Unified Messaging and a series of business-specific tools can benefit from 18.5.3 Network Expansion the strength of a distributed network design without high cost.

Extended networks demand high-grade network functionality relating to 18.7 Résumé Layer 3 switching. These corporate networks can be used through a common infrastructure for performing all business applications (voice, data and video). The A network has to create a stable basis for the wide spectrum of applications support of multiple protocols (as well as Multicast support) and a larger number and services providing strategic benefits and promoting business success. The of VLANs than usual are often necessary. new switched networks are meeting challenges in their development in respect Availability, resilience and redundancy are extremely important for extended of scalability, extended QoS and security, simple administration and minimized networks, because outage even of short duration can entail substantial loss of operating costs. business. High performance is an important point, encompassing the capability By virtue of the distribution of network intelligence to switches of the next of supporting complex applications, e.g. video streaming, CRM, ERP and Unified generation, wireless access points and further intelligent devices in the edge Communication. sector of the network, you can build scalable, administrable networks permitting The possibilities arising from the use of extended networks at multiple sites and cost-effective growth, without compromising performance or availability. from the utilization by many remote users reinforce the demand for a secure 3Com Corporation has been developing innovative networks for over two and resilient VPN service. Support of multiprotocol routing and the possibility of decades and always comes up with entirely novel solutions, allowing you to meet controlling the flow of data traffic by prioritisation, classification and access lists the demand for new services and functions driving productivity, to achieve better are important factors. cost control and reduce administration complexity.

376 transtec = ttec in the Netherlands and Belgium 19. the control of six degrees of freedom. To meet the demands of such applications, such freedom. of of degreesTodemands six the of controlmeet the freedom. of degrees two only of control simultaneous the allow they as applications, 3-D for unsuitable are trackball, or mouse a as such devices, input Standard space. desk less needs but mouse, a as functions same the performs which mouse, upside-down an is trackball a essence, buttons. the pressing top,and on located ball the rotatingby performed is a problem, a be space desk Should m. 2 about of radius a within works mouse wireless the necessary,receiver is the with contact visual direct no that is solution wirelesstechnology. the of advantage The surface. Today,infraredreplaced has mice cableless for technology wireless M ever.than faster is I and comfortable more even becomes windowsinformation the through Scrolling mouse. the moving without applications various within scroll and zoom can user keys. the this,Withfunction the to addition in rocker key or wheel small a with Logiteche.g.or generations,from mouse current The themselves.users the by functions with assigned be can or used, application the by keys function as used either are top,which on buttons of number a has Additionally,mouse the necessary. longer no is pad mouse a desktop,and every almost on work can Additionally,micecleaned. the be to need longer no overcome,mice been optical now up,has pick parts moving the which grease,and dirt dust, of problem the movements.Since the calculate to uses the desktop,which the from second per images hundredseveral produces (DSP). processorsensor Thesignal digital a and sensor optical an contains chip,which small a is technology this of core the Atmouse. the movementof I ball. computer.a the without towork transmitted models is latest Thewhich data of stream a into it converts ball, the of motion rotary the detects which A 19.2 M used. be can sensors integrated fingerprint with keyboards reader card Forcontrol,accesschip cards,credit readingfor programmablekeyboards freely is alternative readers.Another card multi-functional and readers card magnetic readers, readers,bar-code card chip with keyboards are these Amongst unit. one in functions severalcombine which keyboardscomputer. of variety a areThere the on interface one only occupying while functions severalprovide they that advantage the offer all functions integratedadditional other with Keyboards other.each to relation in positioned correctly are keyboard and desk work seat, the that guaranteeto important is it general, syndrome tunnel (carpal injury of risk the reducing thus wrists, and hands for position e.g.the - keyboards Ergonomic lifetime. shorter a has and delicate is foil,which contact thin a only usage.continual with evenlife service long a ensure which contacts, metal-reed gold havekeyboards rule, a Askeyboard. a of functionality life long- the fordecisive is keys the of stability mechanical The handling.secure and ergonomic as well as reliability functional high life, service long provideto expected is keyboard a office, an in use For normal 19.1 Keyboards graphical objects. objects. graphical 3-D of control motion for device a of those with mouse ordinary an of functions designed.the been combinehave Theymice 3-D special nternet navigation, e.g.with navigation, nternet the detects sensor optical an parts, moveabletraditional the having of nstead ice without restrictive cable connections are free to move on the desktop the on moveto free are connections cable restrictive without ice mouse

I ice and Trackballs nput essentially consists of a rolling ball and sensor mechanics/electronics, sensor and ball rolling a of consists essentially M C modelling, requiremodelling, 3-D and reality virtual as such applications, 3-D any TS or tendovaginitis). The result is complaint-free effective work. work. complaint-freetendovaginitis).effective is or resultTS The

D evices M icrosoft’s M icrosoft icrosoft trackball

I nternet atural Keyboard - ensure a healthy a ensure - Keyboard Natural I is a good solution, as the input the as solution, good a is n contrast, low-cost keyboards use keyboards low-cost contrast, n

xplorer or Explorer M icrosoft are equipped are icrosoft cards and bar codes. bar and cards EC Netscape transtec = ttec in the the in ttec = transtec

Navigator, DSP I n I n N etherlands and Belgium and etherlands used to secure performance including big-scale enlargements without any loss in loss any without enlargements big-scale including performance secure to used required.is resolution dpi 2400 a dpi 600 the scans.resolution dpi 2400 a and speeds top at scans quality high for sensor speed. high and The scanning.for standard the become has three of one to on directed is beam each then and (RGB), colours three the into it splits which prism a through light of beam the Secondly,filters.passing by of means differentiationbycolour RGB direct a a onto light of beam white ways.Firstly,a two directing by (three-pass). (RGB) blue and greenred, colours primary the of each for once or pass), (one once document original the D computer.the to transmitted and converter digital analogue/ an by information digital intotranslated then are voltages analogue The values.voltage into these transforms and colour differenceseach forbrightness lamp.a by The underneath illuminated being is it while document motor,original stepping the overa movesby driven device), coupled light-sensitive A principle: following the accordingto operate drawings.Scanners or photographsdocuments, hand-written texts, printing of form the in data importing for aid important an becomehave Scanners 19.3 Scanners process.scanner the controls scanner’sthe to Twainconnection and the driver up sets protocolthat special a Twainsupport. an includes programmewith anyTwain by accessed a be Twainwith can scanner driver any that application. means single This a within from used be can manufacturers differentfrom scanners that so manufacturers peripheral leading bycreated Twain( Twain present.at common are depths colour 48-bit possibilities. colour million 16.7 reaches figure scanner,24-bit 8). this a Withof power the to (2 colour or grey of shades 256 differentiatingof capable is depth colour 8-bit with scanner a hand, other the white. and black between distinguish only can instance,for scanner 1-bit A process.scanning the recorded during colours or grey of shades of number the the is scanner a ratingfor criterion Another printout. the of quality improvesthe which informationpixel additional of means by calculated be to dots recognisedtwo between values intermediate further allows Software specified. always is scanner a of resolutioninterpolated so-called calculated mathematically scanner.a of resolutionphysical the above,refersto described as volume.Resolution data the of quadrupling in results however,resolutionNote, the doubling that detail. higher of image printed a permits resolution higher a that pixels.40,000 reachesillustrates figureThis this dpi, 200 at inch, square per pixels pixels.100 into divided is inch every that means this instance,for dpi 100 recordedwith is image (1”sensors cm). the 2.54 recordedby = inch per pixels of number the to refersfigureimage. printeddpi Theof quality the better the resolution, the higher resolution scanner.a selecting Thein considered be to criteria of number a areThere scan. the of quality epending on the scanning method used, the used, method scanning the on epending enables high resolution high enables HP by further exclusivelydevelopedCCDtechnology has become the most accepted software interface standard for scanners.for standard interface software accepted most the become has T oolkit oolkit of a scanner is measured in dpi (dots per inch). As a general rule, the rule, general a As inch). per (dots dpi in measured is scanner a of

is then used. For films, negatives and enlargements,however,and Fornegativesfilms,used. then CCDis I t determines the actual scanning of the pixels.the of scanning actual the determines t w I n terms of surface area, the scanner provides 100 x 100 = 10,000 = 100 x 100 provides scanner the area, surface of terms n ithout I I t consists of two sensors ( sensors two of consists t n most cases, no more than 600 dpi resolution is needed, where needed, is resolution dpi 600 than more no cases, most n

a n i mportant mportant API

interface ( interface I n such cases the 2400 dpi 2400 the cases such n interpolation n ame) is a software interface standard interface software a is ame) ne-pass scanning can be carried out in out carried be can scanning One-pass

A pplication CCD colour or bitdepth or colour S. The first method mentioned CCDmethod S. first The read head either passes over passes either head CCDread ) in one: a linear 600 dpi 600 linear a one: in ) . This provides the computer with computer . the providesThis P element (charge- CCDelement rogramming element detects the detects CCDelement sensor for high for CCDsensor which performs performs CCDwhich sensor is then is CCDsensor I n addition, a addition, n . I t determines t

I nterface) and nterface) CCD I

f an f On 377

compendium Twain emphasises the need for powerful scanning software. The scanner provides Bar-code readers connected between the keyboard and the computer or display digital data to the computer, representing only what it sees. What the computer terminal via the keyboard lead are in widespread use. Such bar-code readers does with this and how the data is processed further, depends on the software. require neither an additional interface in the computer nor a separate power A particular problem is character or text recognition. Scanned-in text must first supply. Specific programming or changes in the hardware or software are not be processed with a special text recognition (OCR) programme, so that it can be necessary, because the bar-code reader data is treated just like keyboard input. further processed as a text file. The computer does not distinguish between data from the bar code reader and data from the keyboard. Switching between the two input modes is performed Usually, a standard software package is supplied together with scanners. Note, automatically. however, that these scan software packages can differ considerably in the performance they offer. Often only a Limited Edition of the software is supplied. The prices of wireless solutions have dropped drastically recently. Their use It is evident that the software is a decisive factor in user-friendliness. With the can considerably reduce the time and effort involved, particularly for handling HP scan software and the “copy, e-mail, fax or archive” operating buttons, HP awkward machines (e. g. palettes of monitors). The wireless base station is offers a simple solution for reading extensive texts into the PC and for processing installed between the keyboard and the computer and bar codes are recorded them further. Images scanned into presentations and letters also make them more in a way analogous to that of the conventional reader units. Major benefit is the attractive - without involving effort or complicated handling of the software and mobility, since wireless scanners have a maximum range of operation of about 30 the device. m. The various bar-code readers differ in terms of the number of readable codes, A further important point for scanners is whether there is a transmitted light in terms of their code- recognition function (automatic or adjustable), reading option for scanning in transparent media (e.g. slides, films). As a rule, distance and reading mode. Available bar-code readers include pen, CCD and a transmitted light accessory with fluorescent lamps is mounted on the laser gun designs. Reading pens are ideally suitable for small data volumes. A red scanner. LED serves as a light source. In order to recognise 4 codes correctly, the pen must be brought into direct contact with the code that needs to be read. 19.4 Bar-Code Readers A CCD bar-code reader is a near-contact device. The reading distance is between Nowadays, bar codes can be found in every section of the merchandise economy. direct contact and a few centimetres. Laser bar-code readers are capable of They permit small quantities of data to be imported quickly, easily, and above all reproducing codes from large distances (70 cm) and from uneven or curved reliably, into computers for further processing. surfaces. These readers employ a laser diode as a light source.

There is a large variety of bar codes in the various application fields. The codes appear as a sequence of narrow and wide bars or gaps. By optical sensing, the different intensities reflected from the dark bars and light gaps are converted in the receiver into corresponding electronic pulses. These are then translated into computer readable characters by a micro processor, and then transmitted.

378 transtec = ttec in the Netherlands and Belgium 20. Data C either.users of number the to limit no is there And possible. is devices serial via accessremote for posed are distance on limitations no today: technology edge cutting the forms centre.computer the from The metres hundredfew a of radius a within cabling analogue through only possible successively.increased be However,to users and is access servers of number the so-called the Heredesks. operator further adding forpotential restricted the by and managed be to servers of number the tohowever, limited,weresystemsview These a with downtime. server for loss economic the reduceto all,above and,costs, operating cut centres,to computer the in space more gain workstations,to display superfluous saveto centrescomputermany helped already has technology,analogue,This originally function. longer no systemsoperating protocolsor network computers.his restart or diagnostics system out carry settings, BIOS define can administratorsystem a where from workstation,central a to transmitted and computer the at directly tapped just are signals mouse and video keyboard, the purpose, Forthis unit. individual each of front in directly sitting were one if workstations,as multiple or one only via monitored and controlled be can every of function The M (Keyboard,20.1 KV Video, M matrix switches matrix

connections. Administration of any number of servers and servers of number any Administrationof connections. IP switch M KV ommunication provide the solution: Their modular design allows design solution: modular Theirthe provide Digital is based on a simple principle: Several servers servers Several principle: simple a on based is ouse) Switching Data C entre M anagement overanagement

transtec = ttec in the the in ttec = transtec IP

E ven if ven N etherlands and Belgium and etherlands Data C entre M anagement over IP 379

compendium 21. Terminals

PCs or workstations are permanent fixtures in today‘s workplaces. The use of systems, as long as these are interconnected via a network. It is important for display terminals, which are connected either directly or through a terminal server and client to use the same network protocol for communication; nowadays, server to the computer via the serial interface, is in decline. Their use is limited this is normally TCP/IP. to applications which do not require any graphics, which do not place high demands on the computer and where the cost per workplace is the decisive Compared with workstations, X Window terminals represent a cost-effective factor. The main competitor of these terminals, particularly in the Unix domain, alternative since system administration is centralised, which dramatically are powerful workstations or PCs which can process data locally. However, these reduces the effort required for system management. However, potential are significantly more expensive and require more intense maintenance efforts. network bottlenecks should be avoided when using X Window terminals, e.g. In terms of their functions and connectability, X Window terminals rank between by implementing fast backbones. The performance of X Window terminals is display terminals and workstations. calibrated through a standardised benchmark and specified in so-called XStones. Network Computers (NC) have been available on the market for a long time. The storage volume needed differs according to the number of open windows However, they have not yet been properly accepted due to the fact that an and graphics applications, therefore no general specifications can be made in this alternative is available, the so-called Windows-Based Terminals (WBTs). respect. At first glance, both NC and WBT appear to be a step backwards, since, as with Some developments attempt to distinctly increase the performance and area X terminals, data is stored centrally and applications are presented via a uniform of usage of X Window terminals. Given the enormous amount of business user interface. On closer inspection, the advantages with respect to decentralised and technical software for Microsoft operating systems, efforts are underway programme processing become obvious. to integrate these applications on X terminals. The X terminal is then able to The selection of which type of unit and system to be used is especially important simultaneously process and display Windows applications from a Windows because of the high initial purchase price and the subsequent costs for technical server at the same time as the X applications from a Unix server. It is even possible support, maintenance or energy consumption. to exchange information between X and Windows applications with the copy & paste function. For this purpose, there are a variety of software solutions on 21.1 Alpha Terminals the market with very different functions and system requirements. The Xware software, mentioned above, represents a different approach by adding the X Alpha terminals are best used in situations, where graphics capability is not Window functions to PCs. necessary. This means, for example, when the terminal is employed only for database registry. To communicate with the connected computer, the respective 21.3 PCs and Workstations terminals only need to support the required protocols. For historical reasons, different manufacturers have developed specific protocols and terminals for As an alternative to display terminals, PCs with terminal emulation are often used communicating with the computer in their system environments. Alpha terminals today when the user wants to have access to applications on a host computer. are available in single and multi-session versions. The advantage is that smaller tasks, such as word processing, can be performed without overloading the central computer, and the PCs can be used for other 21.2 X Window Terminals tasks as well. The disadvantage of such a solution is the significant increase of the expense of system administration since part of the data is only stored locally, so X Window terminals were developed in order to provide a uniform user interface updates must be carried out separately for each computer. The effort required for for different computer platforms. The need for such a window system, which system maintenance increases substantially in relation to the number of PCs that can distribute information over the network independent of hardware or can be connected to a network. manufacturer, arose as early as the 1980s at MIT (Massachusetts Institute of Another disadvantage of PCs is the greater user effort involved in learning to Technology) in Boston. Based on these requirements, the X Window system was handle the central computer in addition to the PC. This disadvantage can be developed, and was then distributed free of charge to third parties with version compensated by using workstations with the same CPU architecture or at least 10 (X10) in 1986. Many manufacturers seized the opportunity and supported the the same operating system, as the host computer. Examples of these are the current version X11R6 as the graphics and window protocol for their products. SPARCserver and SPARCstations using Solaris. These workstations can ease the However, the classical X Window terminals are being gradually replaced by Thin load on the host computer by providing the possibility of local processing. In Client solutions. This solution expands PCs or NetPCs with X Window terminal contrast to the X Window terminals, workstations can provide local disk space. functions by using corresponding software (Exceed from Hummingbird). Xware The advantage with respect to PCs arises from their uniform use and simplified applications can be installed under Windows 3.1/95/98/ME and Windows NT/ system administration. On the other hand, the disadvantage still remains that the 2000/XP. common personal applications, such as Office packets, are difficult to use.

The X Window system is used to enable easy communication between systems 21.4 Windows-Based Terminals (WBTs) and applications running under different architectures and operating systems. Through the graphical user interface, it is possible to call up several applications A Windows-Based Terminal (WBT) is a Thin Client hardware with Windows on any X Window terminal or Thin Client with Xware software without having CE.NET as the operating system, which is linked to a Windows Terminal Server (for to exit running applications. In this way, information can be imported from one Windows 2000 or Windows NT 4.0) or a MetaFrame Server. A WBT uses either application to another (copy and paste function), which is particularly useful for the Remote Desktop Protocol (RDP, previously referred to as T.SHARE) developed office workplaces with changing applications. Through the X Window system, the by Microsoft, or the Independent Computing Architecture protocol (ICA) from network appears to the user as a single, large computer system. Citrix for communicating with the server. Both protocols are generally supported.

The X Window terminal concept is based on the client-server model. In this An interesting characteristic of Windows-Based Terminals is that they are, as the case, the X Window terminal is referred to as a server, whereas the client is name indicates, completely adjusted to the features in the Windows platform. an X application, that e.g. runs on the processor of a UNIX host computer. The Thus, a multi-user version of Windows 2000 or Windows NT is essential for server, as the displaying part, is responsible for displaying and processing inputs operation, and they cannot co-operate with other systems. This interaction, e.g. and outputs of the screen, keyboard, and mouse. The actual application runs on with Java applications or special company software, has to take place with the the functional part, that is, the client. The client and server can run on separate help of a browser.

380 transtec = ttec in the Netherlands and Belgium other things: other among decision. means purchasingThis a beforemaking awareof be should a tocompared When - - - High-performance - features:following the by characterised particularly is which upgraded an requires longer no architecture this in client A that mean not the in common are Terminals),protocolsthat the but IC the and intranets in use for designed are They applet. Java a as client the on also and server the on run be technology.can Applications A C 21.5 Network IC technologies.incompatible have otherwise would that platforms of sorts all on work to you allow They server.the on run is that application the with interact to it allows and interface user application the (the component architecture. client/server The a in required be normally would that bandwidth network the of tenthapproximately one up take device. end the Withthe to back server the from answer the sends and server the to user the from clicks mouse and entries keyboard transmitsdevice, end the to transported is interface user protocolcomponent interface. Theuser application the from decoupled is logic application the that the and run is application the that ensures server Terminal Server family, server Windowssupported: Windows2000 aresystems operating system. operating following Themulti-user respective the with closely works The IC computers,based sides. client Theand server the on both Server to WinFramethe product product. successorWindows Theextends Terminal non-Windows-basedTohardware,with work also licensed. additionally be tohave not do and license a such haveautomatically with Systemsclient.Windows 2000 ( AccessLicense this. of part important very a are Licenses theft. data as well as loss, data against security with server.the on centrally done is backup data even server; the on using run WBT,When are applications all s are ready to be used in the intranet the in used be to ready are NCs disks hard no need NCs capability audio and graphics Extensive protocols that are protocolsthat Windows(see platform-specificRDP Windows Based and A http://www.citrix.com/download free-of-charge: downloaded be can clients A components: of types important three contains technology A network server component server

s are not suitable for suitable not are NCs C E IC AL). dition, HP dition, omputers (NCs) M A client) that is installed on the end device is used to display to used is device end the on installed is that client) A IC A processors RISC or CISC refers to a referstoThin : This component ensures that the decoupled application decoupled the that ensures: component This , the PC, I Cs, A clients are available for almost any type of end device. end of type any almost foravailable are clients A n addition to this, terminal services services terminal this, to addition n : This software component that is installed on the server server the on installed is that component : software This ata is never available locally.availablenever is DataTherefore, WBTsyouprovide X computers and computers UNIX

-UX,

offers significant advantages, which the user the which advantages,significantoffers NC X, Sun Solaris. The operating system on the on systemSolaris. operating Sun The AIX, IBM C itrix itrix P lient system, that uses Java-based uses that system, Client rofessionaland Windows XP IC IC A technology, the applications only technology,applications A the environments.RDP and A

E very very

I IC client component client IC nternet. They do not use the use not nternet. do They Client A WBTs, just to name a few.A a WBTs,name to just A protocol also supports Java- supports protocolalso A IC C itrix offers the offers itrix

A server component ensures component server A I first needs a Server Server a needs first nternet. However,nternet. does this C AL is needed on the on needed is AL , but simply the simply but PC,

transtec = ttec in the the in ttec = transtec

: The P M rofessional

etaFrame IC A T 4.0 NT Client

NC, N etherlands and Belgium and etherlands etwork PCs 21.6 Network features. additional requirements,however,and different equipmentoffer may they Station), (Network like manufacturers from NCs on informationFurtherT profile.the updating regularly forresponsible also is organisationinternational by adopted been has profile the meantime, the requirementsan minimum the exactly defines Sun and Oracle A “Network C of advantageThin general denominator: common one toreduced be Tocan items these summarize, costs.Lowtraining - prices economical at availableare software - client Webbrowsersas server the onto loaded be only must modules relevant the sincemaintenance and installation time-consuming without running and up be can versions software - latest The companyentire the - requiredare functions desk - backup) data central of terms (e.g.,in higher distinctly is system network-oriented the of integrity and Security - Farcosts administration- lower with respect to ease of application and operation. and application of ease to respect with user the of needs the meet to computer the for is aim the technology,instead but I every nearly fromcontrolled be can electronics home entire the that means user.the of site Thisthe from independentlycontrolled be can unit each that so together networked been have units Fordevelopment.household example, this of results partial interesting certain appliances. been the alreadyhaveThere I units. Theordinary from units communicating and independentcreate to is the by influenced be will as well that is certain is that thing market. only Thethe on prevaileventually will products which and manufacturers which predict to possible not is it yet, of As between positioned category computer area. product this forrealised fully been yet not has administration and management data central of concept the ownership. So of cost higher and client the for price purchase initial high a extensive.in results very This remain disks) hard and drives (e.g.,local equipment the of systemor operating WBTs the that are on based be will (NCD processors I nternet terminal. Therefore, the emphasis is no longer placed on the underlying underlying the on placed longer terminal.no nternet is Therefore,emphasis the between exchanged be can data and together networked are appliances nternet to contrast n C M ost of ost entral management of applications and thus, always uniform versions across versions uniformalways thus, and applications of management entral aintenance of clients or client software is no longer necessary; far less far necessary; longer no is software client or clients of aintenance s and WBTs will have a niche to fill and more and more applications more and more and fill to and niche WBTsNCs a have will O

wnership (T wnership

C omputer Referenceomputer s, which are equipped primarily with primarily equipped are which NCs, , I IBM, ntel eoware (Workstation) or Sun (JavaStation) fully meet these meet fully (JavaStation) Sun or Neoware(Workstation)

CO P Neoware),or entium ). G and its members is available at www.opengroup.de.at available is members its and OG I

nternet. The aim when developing when nternet. aim The lients is the significant reduction of the Totalof reduction significant the is Clients

(NCD NCD s, rather, represent an additional workplace rather,workplace NetPCs, additional representan

CPU

P rofile”Apple,by developed s. The main differences with respect to respect differencess.with main The M processors (Sun), the (Sun), processorsicroSparc- II NC s and ThinPCs ), Tektronix( Netstation), s and the typical hardware typical the and NetPCs The

pen Group (T Group Open I Clients. ntel i960 and i960 ntel I nternet appliances nternet IBM,

must meet. must NC Netscape, OG) IBM

P owerPC NetPCs . This help PC s as PCs The NCs I n 381

compendium 22. Output Devices

22.1 Monitors sufficient speed. In TFT displays and tube monitors similar effects are encountered to a significantly smaller extent. (For picture tubes a compromise always has to 22.1.1 Units of Measurement and Terms be sought between rapid response time and the suppr ession of flickering by sufficient after-glow of the phosphor.) The resolution of a monitor refers to the number of pixels that are displayed on the monitor. One picture element, or pixel, represents the smallest unit that can 22.1.2 Types of Monitors be addressed. 22.1.2.1 CRT Monitors The refresh rate specifies how often the entire picture is redrawn per second. The higher the rate, the more stable the picture appears. Optimum rates start 22.1.2.1.1 Principle of operation at about 73 Hz at which the eye cannot detect a flicker thus limiting the eye strain - especially during relatively long sessions in front of the monitor. The In CRT (Cathode Ray Tube) monitors the image displayed on the monitor is recommended guideline value nowadays is 85 Hz. produced when an electron beam hits a phosphor coating on the inside of the screen. To create the colours, three phosphor coatings for the primary colours The line frequency refers to the time required to draw a monitor line. It is red, green and blue are provided. The colour mixture is based on the so-called calculated from the refresh rate and the number of lines for the particular video additive method, i.e., when a beam hits one of the phosphor coatings, light of the mode and is measured in kHz. It therefore depends on the resolution and the corresponding colour is emitted. Thus, by combining the three primary colours, refresh rate of the monitor. other colours of varying brightness can be created.

The video bandwidth reflects the quality of the video amplifier. If the video To ensure that the rays hit the intended phosphor dots, a mask is located in front bandwidth is insufficient, shadowing may occur in the case of light vertical lines of the phosphor coating. against a black background. At present there are three types of picture tubes on the market, which are The brightness control adjusts the black level of the picture. (The contrast differentiated by the kind of mask used: adjusts the brightness level). If the brightness level is set too high, black elements appear grey. 22.1.2.1.2 Shadow mask, slotted mask and aperture grille

The contrast: The contrast control is used to adjust the brightness wanted. The In the shadow mask the mask consist in principle of a perforated metal or contrast should be adjusted so that the picture is sufficiently bright, but is not ceramic sheet with round holes through which the electron beams are directed so of uncomfortable intensity, as is evidenced by the righthand image edges being as to strike the correct dot on the screen. The shortest distance between the two distorted in colour. dots of the same colour on the screen provides information about the fineness of the shadow mask and indirectly indicates the number of dots a monitor can display The focus refers to the shape of the electron beam which, when it strikes the as single elements. The smaller the shadow mask the grater the definition of the phosphor, should be as close as possible to perfectly round over the entire image. monitor area for all resolutions and frequencies. If the focus is poor, the displayed In the slotted mask, however, the holes are rectangular. This type of monitor characters and lines become blurred, particularly in the screen corners. It should is no longer prevalent in the monitor sector, this technology is used only by be borne in mind that the picture definition is also influenced by the video manufactures of television sets. bandwidth, the convergence and the selected screen resolution. The aperture grille, on the other hand, consist only of vertical wires under high tension, through which the electron is beam is directed. Convergence designates the alignment of the three electron beams for red, green and blue. Ideally these beams are directly one on top of the other in The advantages and disadvantages of the different masks types: order that a clean, white dot can be produced. If this is not the case, lines and The shadow and slotted mask picture tubes are less sensitive to vibration due characters are displayed with coloured edges and lack definition. to the perforated sheet, but as yet cannot offer the contrast and vibrant colours of an aperture grille. In addition the enlarged surface means the are more subject Moiré refers to a wave-like pattern running over the screen, which changes when to so-called doming, the effect resulting from intensified heating by the electron the picture width and height is altered and only happens with certain resolutions. beam and leading to deformation of the mask. The electron beam is hence unable These rippled images are caused by interferences between the video signal and to traverse the shadow mask without problems, resulting in distorted colours the monitor mask and occur particularly with patterns of lines or dots. Ironically, especially in the corners of the screen, and poor distribution of brightness. Shadow the better the focus, the worse the moiré is likely to be. Patterns of lines or dots, mask monitors are suited particularly for use in industrial environments due to extending over the entire screen, are particularly suited for testing purposes. being insensitive to vibration. By virtue of its design, the aperture gille converts less of the energy into the The picture stability: Keeping the high voltage of a monitor stable under electron beams into heat, meaning it is less susceptible to colour distortion and the wide variety of operating conditions demands high technical input. Poor can achieve higher contrast and more vibrant colours. The downside is, that it stabilisation manifests itself particularly in bright/dark changes. A weak high is somewhat more sensitive to vibration and is stabilised by two damper wires voltage stabilisation then leads to constant changes in picture width, so-called perceptible to the trained eyes as slight shadow lines. “pumping”. Another advantage of a Trinitron Monitor (Trinitron is a registered trademark of The smearing effect is the phenomenon when a moving object on the screen Sony Co.) is that it has a flatter screen surface. The basic form of conventional no longer has any clear definition, but has a “tail” in its wake. This effect is mask systems is spherical and is difficult to flatten, even for the modern FSTs (Flat encountered particularly in passive LCDs (STN, DSTN) and is conditioned by Screen Tubes). In contrast, an aperture grille monitor is shaped like a cylinder, the the slowness of the liquid crystals used. For fast, moving pictures with high screen surface of the Trinitron Monitor is considerably flatter. The flatter screen differences in contrast, the liquid crystals cannot assume the new condition with surface prevents images from being distorted. Straight lines appear straight, thus

382 transtec = ttec in the Netherlands and Belgium in moving images asaresult oftheslowness ofthecrystals. areeffect, used, slow liquidcrystals whichhave thedrawback effects ofsmearing fieldandtheythenapplying anelectric „lose“thisalignment again. To reduce this lines are addressed lineby line, asinamonitor. are alignedThe by crystals briefly Passive matrix - Uneven illumination dueto thebacklight withmoving- Smearing images ordeadpixels- Missing expected, dependingonthetechnology used: red, green andbluepixels isfixed.Instead, the following have restrictions to be dot consisting ofthreeor lackofdefinition,becausethearrangement ofevery free from misalignment any inconvergence errors distortion, suchasgeometric consequenceIn ofthetechnology employed, thedigital LCD displays are ofcourse transistors are required). with aresolution of1024by 768,for example, 1024x7683=~2.36million costs and hence theunitprice are stillrather high(inorder to produce a display chips. ofsemiconductor production to Due rates, thehighreject theproduction effect.smearing These thinfilm transistors are produced ina process similar tothe they are to re-addressed, allowingbeusedandavoiding faster liquidcrystals the by way oftransistors hastheadvantage that thecells maintain theircondition until pixels againstbeingaffected by adjacent pixels and prevent crosstalk. The control whichalsoprotect the cell.(TFT), As aruletheseswitches are thinfilmtransistors switches, oneeachfor thecolours red, green andblue, are usedto control the TFT displays isactive. This signifies that singledot, three electronic for every displays, toContrary thepassive liquidcrystal theaddressing inthe control ofeachsinglepixel, asaddressing. known The essential differencetwo between the types of available LCDs isthebrightness 22.1.2.2.2 Active andpassive matrixLCDs There are ofLC two types displays: by amatrixofpixels. the colours ofred, green orblueproduced by colour filters. The imageisgenerated asawhole.the picture To obtainacolour display, three pixels perdotare usedin grey display. levels for ahigh-quality The screen issubdividedinto pixels composing thiswayIn thebrightnesscanbecontrolled numberof to obtainthenecessary display. of thelight ischangedand, accordingly, ofthelight canpassthrough the onlypart andhence fieldisapplied,twisting angleofthecrystals the anelectric If crystals. angle to thefirstoneandcanpassthroughtwisting due by theliquid to the molecules. This light thenimpingesasecond polarisationfilter arranged rightat fieldisapplied,twisted thepolarisationischangedalong As longasnoelectric and isaligned by polarisationfilters. displays. The light isproduced by light sources at therear (backlight) ofthedisplay areThese usedinLCDs properties to control thepassageoflight through the through. are fieldisapplied, aligned accordingly. thecrystals When anelectric molecules. ofthemoleculesaltersThe thepolarisationoflight twist passing elements, theso-called ofagroup ofdisplayThis makesuseofthespecialproperties type ofchemical 22.1.2.2.1 Principle ofoperation 22.1.2.2 LCD Monitors light assunraysextraneous andlamplight. the blacktinted glassofaBlack Trinitron monitor agreat absorbs dealofthe itsuitablefor andgraphics CAD/CAM modern applications.making Furthermore,

(STN, DSTN): In this type of liquid crystal display, ofliquidcrystal thistype (STN, DSTN):In theindividual l iquid c rystals, whicharerystals, transparent andhave twisted transtec = ttec in the Netherlands and Belgium and Netherlands the in ttec = transtec active matrix active

the various mask types). mask various the of drawbacks(Cf. used.and advantagesbe should masks shadow with monitors avoided,CRT be cannot vibrations and shocks environmentwhere industrial an In for.called are brilliance and colour true absolutely which in applications for and CAD/CAM in applications, graphics-oriented in particularly displays panel flat of instead grille aperture with monitor CRT a for go to advisable is It quality. picture resultant the fordecisive is method interpolation the pixels.of eliminating quality or The dpi. 1200 x 1600 supports physically which resolution, a such with monitor a buy to thereforeadvisable is quality.It in involve loss does extrapolation howeverevery duration, short of applications for suitable means all by are extrapolation, through resolution higher a achieving size, this of displays Flatpanel standard.already is dpi 1200 x 1600 of resolution a displays 20-inch Forapplications.intensivethe space- environmentforoffice normal a in prevalentmore ever becoming are displays 20-inch and 19-inch Larger these. as such cases in suitable arepixels 1024 x 1280 of resolution standard a havingtoday displays panel flat 17-inch display.panel flat 15-inch a of limits the reach may user the time, same the at opened be to arewindows monitor.or applications severalHowever when cm) (38.1 15-inch a TFT using when resolutionpixel 768 x 1024 minimum a time. demand buttons,Today’s and menus numerous their with applications, the all better and cheaper getting are displays,which panel flat to monitors the determining for important is used is monitor a which for applications of type The 22.1.3.1 Monitorsizes andapplication 22.1.3 MonitorRequirements 85 Hz). Further innovations include: a lower power consumption of 15 watts forwatts 15 of consumptionpower lower a include: innovationsFurther Hz). 85 min. resolution, 768 x 1024 with monitor 17-inch (e.g.a device the of size and resolution the on depending determined been has raterefresh minimum a time, first Forergonomics.the and quality of terms requirementsin and guidelines stringent 1999. new Toin the passed meet must be monitors TCO’99-certified, The emission. low as referredto usually arerecommendation this by stated values the below radiation with Monitors 1990. in institute inspection Swedish the by specified was charging,which electrostatic and fields electric alternating fields, electromagnetic alternating to respect with levels radiation the concerning the is standard important levels.An radiation to regardsdifferent with institutes,by defined thereforehavebeen Guidelines emits. it radiation of intensity the is monitor a selecting when factor important Another 22.1.3.3 Radiation transtec.fromavailable is this on informationFurther conditions. viewing and posture working adjustment, seating the are instrumental Also health. preserving and relaxedwork permitting set-up:Workplace facility.tilt-and-swivel reflection, and interference from freesize, convergence character focus,adjustable control,sufficient dynamic definition, are: factors Decisive and TCO’99/TCO’03label Angel seals. requirement: Minimum Monitors: forRequirements Ergonomic 22.1.3.2 Ergonomics available.made be should 1200 x 1600 of resolution a area, this in applications complexFor suitable.more arepixels 1024 x 1280 of resolution a with monitors cm) (53/55 21-inch or cm) (50.8 20-inch cm), (48/50 19-inch DTPapplications, and FordesignCAD, graphic screen diagonal screen TCO’99 Inter-/Extrapolation: guideline, based on the TCO’92on and based guideline,TCO’95 precursors,was The right workplace set-up also plays an important part in part important an plays also set-up workplace right The . There is a clear trend away from conventionalfromawaytube trend clear a . is There size and picture quality, flickerfree, sufficient contrast, good contrast,sufficient quality, flickerfree, picture and size MPR II (tested at TÜV Rheinland), environmentalBlue at Rheinland), (tested TÜV II MPR

is a method of enlarging or reducing images by adding by images reducing or enlarging of method a is

MPR-II , a recommendation a ,

383

compendium the energy-saving mode (previously 30 watts), a ban on substances containing 22.2.2 Pros and Cons of the two technologies bromine and chlorine, the avoidance of heavy metals and mandatory recycling regulations in the relevant countries. A high degree of saturation of the colours and uniform illumination are the advantages of the dependable LCD technology. Devices incorporating LCD TCO’03, in force since the end of 2002/beginning of 2003, is based on the technology also have the advantage of economical purchasing costs. TCO’99 standard and makes more stringent demands on the two areas of tube True colours and the absence of any pixel errors are clear advantages of the monitors and flat screen displays, defining, for example, vertical adjustability, DLP devices which, while being slightly more expensive, outshine alternatives by cable, enclosure and recycling requirements. bringing more light to screen and providing very good picture quality.

22.1.4 Graphic Standards 22.3 Printers

In the PC area, different graphics standards have been developed over time Inkjet printers and laser printers currently take the forefront, while matrix printers to ensure that graphics cards, monitors and application programmes are all and thermo printers carry hardly any weight in everyday life. compatible with one another. Video Graphics Array Standard (VGA Standard) and Super VGA Standard were the first to become prevalent. The resolution of 22.3.1 The Technologies the VGA standard is 640 x 480 pixels; 16 out of a palette of 256 colours can be displayed at the same time. In an additional mode 256 colours can be displayed, 22.3.1.1 Inkjet printers but only with 320 x 200 pixels. Various technologies have been developed for inkjet printers. Distinction is made Super-VGA (SVGA) and XGA stand for higher resolutions of up to 1024 x 768 between solid ink printers and liquid ink printers, as well as between drop on pixels. In order to display 16 out of 256 colours, a graphics card memory of 512 demand technology and continuous flow technology. KB is required. With a 1 MB memory expansion, a total of 256 colours can be displayed from a range of 16.7 million colours. Thermal Inkjet Printers For workstations, the graphics standards of the PC area are of little importance. This inkjet technology from HP has established itself as the mainstream The graphics cards used with SPARC Stations usually support a resolution of 1152 technology. Printers operating according to this principle are fast, provide x 864 pixels with a refresh rate of 66 or 76 Hz. Graphics cards that support the high print quality and the purchasing and maintenance costs are low. These VGA standard or reach resolutions from 1024 x 768 up to 1600 x 1280 are also printers embody a very robust system in which the printer mechanism and the available in the SPARC area. high-precision print heads are kept as separate units. The entire print head is accommodated in replaceable ink cartridges produced using the same production 22.2 LCD Projectors technology as for microchips. The micro-lithographic production technology provides for economical mass production attended by constant quality. A round, For presentations to larger groups, the screen display may need to be shown to mono-crystalline silicon wafer is the starting material which is etched to provide all participants with the use of a projector. While presentations using an overhead the structures necessary for the ink transport, ink nozzles and, of course, for the projector are time-consuming to prepare (printing transparencies, making copies, electrical connections and components. When an ink cartridge is empty, the new etc.), the use of LCD projectors saves a significant amount of time and changes cartridge automatically comes with a new print head, ensuring years of high print can be made right up to the last minute, even while the presentation is underway. quality at a comparatively low price. What is more, animation can be added, making the presentations interactive and In this printing method, the print head travels on a carriage horizontally over the giving extra interest. . A strip of the image is printed, then the paper moves on ready for the next strip. The printing operation is of non-contact type. Wherever a dot needs The development of the present-day generation of projectors is going in the printing, a tiny ink droplet is fired from one of the nozzles in the print head. direction of handier and quieter devices for use in conference rooms, which are That is why the technology is also known as drop on demand, as opposed to also suited for mobile use. Sound levels of less than 30 dB, ultramodern lamps the continuous flow technology (see below). It ensures that only as much ink is with a service life of up to 6000 hours and a weight of less than 2 kg are already used as is actually required for printing. In the meantime there are highly complex common. The constantly increasing ANSI Lumen figures permit use in rooms operations going on inside. of almost daylight conditions. The projectors from Philips, for example, operate The nozzle firing the ink is a miniscule chamber with a resistor at the bottom, with the so-called Limesco technology (Line Memory Scan Converter), including which becomes hot when power is supplied. The ink exits the reservoir through a chip permitting screen-filling display of higher graphics standards through the capillary force and passes through channels into the nozzle chamber where it is projector’s own graphics resolution. heated to temperatures in excess of 300°C in fractions of a second. The vapour bubble forming at the bottom of the chamber presses the remaining ink in the 22.2.1 The Technologies chamber through the nozzle shaft and out of the opening. At the nozzle opening, which is thinner than half a human hair, the vapour bubble collapses liberating The most widespread technology is LCD technology. Light is split into the shock waves shooting out the ink droplets at a speed of some 100 km/h. Once primary colours of red, green and blue, is conducted through one LC display the ink droplet has left the nozzle, the resulting vacuum in the nozzle chamber each as a light valve and at the end of this process is united again, hence each draws ink from the reservoir to replace the ink that was ejected. The process can individual dot and each one of the three primary colours has an LC display of its repeat itself several thousands of times a second. own. Nowadays sophisticated computer simulation is used for analysing the droplet formation and ejection, enabling on-going further development of the operation. The DLP technology is based on multiple, miniscule, square-shaped mirrors This is important, for instance, with a view to the dynamics of the process, since applied to a DMD chip. The light is reflected to the screen only via tilted mirrors. the print head moves on continuously as printing takes place. The droplet does The human eye cannot detect this high frequency in which the mirrors are not fly along a linear path, but describes an accurately calculated, arcuate path switched on and off. Different activation times permit the display of graduations. towards the paper. These technical parameters are incorporated in the printer The colours are generated by a colour wheel filtering the light into the three control and the entire operation is optimised accordingly. primary colours. The developers’ attention is also levelled to the size of the ink droplets. Although

384 transtec = ttec in the Netherlands and Belgium Printers22.3.1.3 Matrix printers. inkjet of those belowperceptibly are page per volumes.costs print running Thehigh for primarily recommendedare they printers, inkjet of that than higher is printers laser of price purchase the Sinceprinting. during level noise low the as well as capability graphics and quality print high their are printers laser offeredby benefits The software. of irrespective points, adjoining the to relation in dots of position and size the optimising by transitions smoother and edges and lines sharper curves, rounderproduce can as method HP,positioning. dot variable or size this dot controllablereferstoinstance, for through quality print improving of methods developedhave manufacturers Several transparencies.or paper onto step one in image the transfers then drum belt. charged special The a on accumulated is printed be to image the times,four machine the through paper the moving than (C, colour Rather mix Y, K). every forand once M out carried is operation complete the that is printers laser colour available the differencein printers. laser only colourThe and monochromefor both applies method This pressure. high and heat applying byfixed and paper the transferredonto then is toner points. neutralised Thethese on toner the up takes drum process,the printing the of step next the In neutralised. then is drum transfer the on charge electric the laser the of point impact off.the Atbeam laser the switches printer the set, be to not is that dot line Forevery drum. the to line-by-line image the transfers mirrors of system a by guided beam charged.laser a electrically Then is drum transfer the that is process printing the required.in arestepsolvents first The liquid no hence and place take processes chemical no which in process printing physicalprocess, a electro-photographic an use printers Laser principle. page the on based is technology laser - computer the in generated file print a - principle line so-called the on based aretechnologies printer above-stated the Whereas 22.3.1.2 LaserPrinters needs. specific torequirements. answers special are They printing normal meet to enough compact not is equipment the and expensive very is it but printing,colour in quality,especially high very achieves technology paper.the onto stream ink This this from deflected are droplets individual printing, During vessel. collection a into reservoir a from field electrostatic an in continuouslyflows ink which in technology this using operateprinters few a printers inkjet flowContinous carefully. very handled be thereforetohave limited.and delicate is are Theyuse daily for results printed the of suitability the method,transfer thermo the with as Similarly state.melted a in ready is ink respective the and reached been temperaturehas operating the until again on switched is printer the when incurred are times waiting Longsystem. printing the of cleaning time-consuming off,entailing switched is printer the after again printing.cools for ink reservoir The ink an in ready kept is and liquefies it which in colour each forprovided is chamber melting costly.separate so A them makes which technology the is relativelyexpensive.It are printers these of maintenance and price purchase the but good, very is image printers. inkjet print The phase-change called also are printers these solid), - liquid - (solid ink the of state in change the to Due again. cools immediately it where paper the onto squirted is condition melted the in and heatedtemperature, is The increased. is speed print the and paper normal on even high very is quality print the dot, representedper be can colours More arise: benefits following the another.practice, one In of top on dropletsseveral printing by better dosed be to volume ink the allows This smaller.ever getting are droplets picolitres,the 10 of range the in alreadytoday For the first time they allowed text and graphics to be printed on one and the and one on printed be to graphics and text allowed they time first Forthe PC. every almost of vicinity the in found be to was printer matrix a past, the In solid ink printers ink solid

R esolution are used for colour printing. Ink, which is gel-like at room at gel-like is which Ink, printing.colour for used are E nhancement

are becoming ever rarer on the market. Only market. the on rarerever becoming are T echnology (RE echnology T). This built-in feature built-in This T). transtec = ttec in the Netherlands and Belgium and Netherlands the in ttec = transtec of wax films, but the wax is heated only to the point at which it melts and is and melts it which at point the to only heated is wax the but films, wax of The doubtful. is benefit their that professionalmodels the byprovided that of short so falls sector computing home the for offeredespecially versions economical sector.the computing of home quality and Theoffice the for question into come not do printers sublimation thermo that high, so are paper) special films, (colour costs operating mixed.and costsis purchasecolour The which in way the by but resolution, the by not defined is quality print colour that proof further is This dpi. 200 than less even of resolution low comparativelya with output image high-quality their achieve printers these that connection this in interest of is It art. graphic in prints prooffor sufficient quality a in printed be togradations colour fine permit also printers sublimation Thermo paper.the penetrates and vapourises it that hot so becomes paper.on it waxprint The and film carrier a on situated wax coloured heat prints, colour Forhigh-quality market. the on significancegreatany of longer no are machines) fax thermo to (similarly heat by visible paper the in incorporateddyes render printers.which sublimation printers Thermo thermo and transfer thermo printers: thermo term the under subsumed technologies two at look Wewill 22.3.1.4 Thermo Printers printed. be tohave copies) (carbon forms multi-page which in areas other and firms logistics in areas,particularly some in indispensable todayare printers matrix same, the increase.All costs running the speed, operating slow extremely the from Apart one. black the of instead ribbon four-colour or three a using prints colouredproduce also can printers matrix Some comparativelynoisy.are printers matrix attractive, particularly not is dots individual of composedproduce they image print the that fact the from Apart page. same printer. laser a with seconds within performed be can plotter a with processconsuming time- a reasons,is technical for which, output Graphicsdrivers. adjust to having without performed be can that task simple a is HP-GL-HP-GL/2 or supporting printer laser a foroutput. plotter graphicscorrespondingexchanging Thus,a the intotransformed be to and printer laser the byprocessed be to commands plotter enables printer laser a protocolon plotter HP-GL/2 or HP-GL the Implementing quality.print top ensures which language, vector-based a is HP-GL HP-GL. of development further a is PCL HP procotolintegratedin HP-GL/2 (HP HP-GL the manage also units Some (GDI). Interface Direct Graphical the support commands 6 PCL Microsoft the Windows,under because particularly expected be to are performance in increases Significantcommands. compact more and shorter from resulting throughput network higher and quality graphics,improvedprint complex of acceleratedespecially printing,include 6 PCL version. in innovations earlier The the downwardto compatibleis which 5 PCL HP of development further a is 6 PCL HP manufacturers. other numerous by used already is which plotters, especially printers, of control the for HP by developed ( PCL HP structure. command flexible and performance high its to due sectors many in used is PostScriptfont. a define to curves) (Bezier curves mathematical uses that language description page printer-independent a the manage printers laser Most protocol.printer forsynonym a as used often is language Page description 22.3.2 Printer Protocols printing. sublimation thermo to one, economical an alternative,particularly genuine representa they whom for artists, commercial professionalby particularly use.used generalare Theytheir discouraging factor a is consumables the of costs high the segment, price lower the offeredin are expensive.are needed films wax the and used be can films and paper special scores.Only and creases to susceptible are prints the but good sufficiently is renditiondot. colouredcolour Thea as paper the to applied thermo transfer printertransfer thermo P rinter C ommunication

is a simplified type. It also operates with the use the with operates also It type. simplified a is thermo sublimation printers sublimation thermo HP PCL HP L anguage) is a page description language description page a is anguage) and/or G

raphics

PostScript L anguage) plotter language. plotter Theanguage)

E ven though some printers some though ven protocols.is PostScript are offered.are They

385

compendium SS

23. Multimedia

Multimedia has for a longer time not just been a widespread catchword, it has connected video recorder or video camera. Thus, either entire video sequences been implemented in many applications. or freeze frames can be manipulated, stored or incorporated into computer demonstrations as desired. In general, multimedia is a term used to refer to the combination of various However, a is needed before video clips can be created. A video card media, such as still pictures (e.g. those produced by digital cameras), stores video sequences or freeze frames from a connected video recorder, camera audio, motion pictures, texts and graphics. Many computers perform or television set on the hard disk. Video software then offers the possibility of additional tasks. The user can listen to conventional audio CDs, send e-mail manipulating these images. Since the video data can be reproduced and edited messages with embedded sounds or have the computer serve as a telephone without noticeable loss of quality, video editing is reduced to file copy and delete system to forward or answer calls. commands. Furthermore, the computer processing of video images offers the advantage that each individual section of images from the digitised material can Video conference software is becoming increasingly important, since it can do be accessed within seconds, eliminating the time and effort involved in rewinding. a lot more than just transmit normal video pictures. Conferences can be held A variety of video cards with different performance features, described below, are between various connected partners and documents can be exchanged rapidly available for transferring individual video images or entire video sequences to the and cost-effectively over great distances. In the context of multimedia, the PC or workstation. CD, DVD and jukebox media are mentioned as well. This is because the data In order to display real-time videos or television pictures on the monitor without quantities are usually very large, requiring fast and cost-effective high-capacity the complexity of digitising the video data, an overlay card is needed. Overlay storage media. cards usually only have a limited, or no grabbing function at all, and therefore can only be used as television receivers or for surveillance cameras. The video signal is 23.1 Digital Cameras transferred from the overlay card to the graphics card so that it can be displayed without wasting computing time. The video image is displayed in a separate The number of multimedia applications continues to grow and therefore digital window on the monitor. cameras become more important. In order to process pictures from a conventional camera by computer, the film must be developed first and the slide or print In addition to the video and overlay functions, some video cards incorporate a must be scanned. Digital cameras produce a digitised image stored in an internal graphics card. This means that a separate graphics card is not required and only memory which can be transferred to the computer via an interface, a floppy disk, one slot in the PC rack is needed for both functions. Video cards with a TV tuner or a CD/DVD. go one step further. In addition to live videos, they allow any live TV broadcast to The user can enlarge or reduce the pictures, incorporate them into documents, be displayed in a window on the screen. and send them using online services. Desktop publishing, presentations and To display and digitise entire video sequences on-screen, real-time digitisers are relatively simple marketing brochures can be illustrated quickly and easily without required. They convert the video signal into digital picture sequences. the need for a scanner. The cycle “photograph - view - edit - print” can be The main problem in storing video sequences is the enormous amount of data completed if a suitable colour printer, e.g. from HP is also available. generated by the digitisation of pictures and videos. Digitising one video image with a 24-bit colour resolution requires about 1.2 MB. Thus a PAL signal of 25 23.2 Voice Applications images per second generates a data volume of 30 MB per second. Therefore, to store video sequences, a video card with hardware compression should be Like music, voice is an analogue signal which can be converted by digitisation used. Data compression reduces the memory space required, and at the same into a computer processable format, and subsequently reconverted into an time increases the speed at which the video images can be processed. Besides, analogue signal using a sound card. In order to input, output and process sound such compression reduces the bandwidth required for data transmission over the information, the computer needs an audio interface card, called a sound card in network. the computer industry. Compression without degrading quality and without taking too much computing The audio card samples several thousand times per second the analogue audio time is possible only up to a ratio of 1:2. Practically all solutions for video signal input from a microphone or patch cable (sampling). The sampling quality compression must however have far higher compression ratios. Software solutions depends on the resolution of the digital-analogue converter (e.g. 8, 10 or 16 use algorithms that have compression ratios of up to 1:20, with only a slight bit) as well as on the sampling frequency (number of readings per second). This degradation of image quality. Given the limitation of this compression ratio, many process is called Pulse Code Modulation (PCM). To transfer voice signals over video digitisers use the JPEG compression standard. Special integrated circuits on long distances, for example via e-mail, the quantity of data must be kept small. the video card allow images to be compressed and decompressed by factors of For that purpose, the signals are compressed before transmission using ADPCM up to 1:10 in real time, i.e., at 25 frames per second or more. However, when (Adaptive Differential PCM) and decompressed again by the recipient. compressing using this algorithm, there is a risk of losing some information, especially at higher compression ratios, i.e., not only is there a difference between 23.3 Video Processing the original image and the compressed one, but also the former cannot be reconstructed from the latter. So-called cube effects occur and the components of Video cards are applied in a wide range of applications today. Apart from their the image become extremely fuzzy. The degree of compression can be adjusted. use for creating digital video clips for training purposes or business presentations, In practise, a compromise has to be made here, since high quality means little the production of live videos or video-on-demand is arousing interest. Live video compression and therefore large quantities of data. allows the user to follow news broadcasts and run applications at the same time. Motion JPEG only works with single images; in effect, a file is created containing nothing more than JPEG-compressed single images. Another field of application is the real-time transmission of compressed video MPEG goes one step further in that it takes into account that there is usually images via networks. The images can be stored on a video server equipped with only a small difference from one image to the next. This method eliminates the an appropriate card. By means of video-on demand, remote users with network similarities between consecutive images. As few as two images per second contain access to the video server can watch the stored video films on their workstations the full independent information of a single image. All the other intermediate or save them locally for further processing. images contain only the changes to the preceding image. It requires an enormous To create presentations, images can be read into the workstation from a suitably amount of computing power to determine which parts of the image can be used

386 transtec = ttec in the Netherlands and Belgium SS

in the next image. This accounts for the high costs of video digitisation cards with For product presentations in particular, it is obviously suitable to output the MPEG motion compression. previously digitised video sequences onto a video recorder or television after To ensure an appropriate output quality of the digitised video sequences, it is processing. For this, the video card must have a separate video output. The recommended to use only high-quality starting material. Problems with image advantage is that a television set is considerably cheaper than a large screen quality soon become apparent when digitising VHS videos. S-VHS or, even better, display monitor with an appropriate diagonal size. Furthermore, the videos professional Betacam are much more suitable for this task. The fact that image can also be presented to a larger audience by means of a video projector. The quality is lost due to the compression does not mean that the input quality is production of video tapes is another interesting option, especially since the data unimportant. Faults such as drop outs or noise become doubly apparent after stored on the tape can then be displayed on a TV set by any video recorder. digitising. Compression methods such as MPEG are very sensitive, particularly to signal noise, due to the fact that a great deal of information changes from one 23.4 Video Conference Systems image to the next. Costs savings is the main reason why companies value video conference Once the video images are stored on the hard disk, they can be edited and systems more and more. Audio/video communication leads to easier and clearer manipulated as desired using the video-processing software. Thus, by means of understanding between individual or multiple communicating partners. The dissolves and other special effects it is possible to create ones own video films. extensive spread of ISDN has facilitated the implementation of video conference With suitable libraries and supplied programme interfaces, users are able to write systems. Most systems available today, in addition to the required software, their own programmes for image processing on the workstation or PC. Thus, include an appropriate graphics card, a video camera, a microphone and an ISDN for training and business presentation purposes, for example, the desired image card. With ITU industry standards H.323 (LAN), H.320 (ISDN) and T.120 (common sequences can be arranged and combined with text inserts. data access), it is possible to communicate with video conference systems from Tutorial programmes can be designed so that the individual user can choose other manufacturers who meet these standards. which chapter he wants to display on screen and with which degree of difficulty. The programme then searches for the corresponding pictures. compendium

transtec = ttec in the Netherlands and Belgium 387 24. Uninterruptible Power Supply

Uninterruptible Power Supplies provide all electric devices with power from a 24.2.3. Online Technology battery, if the mains power fails. An unexpected network disruption may lead to the loss of data and programmes, and the resulting costs and damage cannot This model provides more data security, but it is generally more expensive than be enormous. The contents of a disk may be damaged if the power failure occur offline or line interactive models, due to its complex technology. The incoming during a write operation or if the contents of the write-cache memory was not mains power is electrically isolated and regulated. The rectified voltage is previously transferred to the disk. This is why it is important to shut down the smoothed, stabilised and transformed back into a clean alternating current. computer in a controlled manner when such disruptions occur. This technology guarantees that the connected devices will be free of voltage fluctuations, disruptions, noise, or spikes. However, power failures are not the only reason for network disruptions, there are The battery is charged during normal operation. Should the power fail completely, a variety of causes for this and and they can occur in many different ways. the charged battery takes over the power supply to the inverter smoothly and without interruption. This technology continuously produces a completely new 24.1 Problems, Causes and Effects voltage signal, independent of that of the mains. So online UPS systems do not operate without activation times, but offer the additional advantage of functioning as voltage stabilisers and line filters. A disadvantage of online UPS is larger energy expenditure due to the continuous supply through the inverter. The high amount of heat which is generated by this necessitates fans, which also leads to more noise. Therefore the optimum area to install this system is in a computer room. Common models also have a bypass function, so that the devices are not just independent from the network, but also from the UPS. In the case that the inverter is defective, the system is bridged and the devices are immediately

supplied with energy from the network. Additionally, the systems have a manual bypass function, a simple flip of a switch allows you to exchange the batteries and perform maintenance work while the system is running without influencing the devices.

A comparative overview:

9 Types of possible voltage problems

24.2 Technologies

24.2.1. Offline Technology Devices are directly supplied by the network via a filter, without voltage or frequency regulation. The devices are supplied by the battery through the inverter after a switch time of < 10 ms. Offline models function like an emergency generator and are only activated when the power has already failed. The switch times listed are generally sufficient for PCs, this time is usually not long enough to cause disruption. In contrast, even this short switch time can lead to data loss in Line Interactive Technology other systems.

24.2.2. Line Interactive Technology

Just as with offline technology, devices are directly supplied by the network via a filter when using line interactive technology, but voltage regulation is possible. A control circuit connected in parallel to the power supply of the UPS regulates fluctuations of the main power within a range which is tolerable for the user and allows full utilisation of line resources because, in the event of a power interruption, the UPS does not immediately switch to battery power. The result is a longer battery life. If the power fails, the devices are supplied by the battery through the inverter after a switch time of < 10 ms. This improved version of offline technology, however, is not sufficient if the connected devices are sensitive to the phase shift caused by the switch from line to battery power. This is the case with many telecommunications systems.

Online Technology

388 transtec = ttec in the Netherlands and Belgium constantly.run tohave that applications critical with use for hours several span that times bridging with availablefailure.areSystems power a of event the in savedproperly be always programmescan and data that time energystored power. battery with time This short a for supplied be only can devices peripheral and computers temporarily, connected system the safeguardsa UPS a Since devices. connected the of all including basis,regular a on testing function WeUPS recommend damaged. be can overloaded,beforeit is it if down system the shuts that mechanism self-protection a containensured.systems be UPS Thelonger no can system the of shutdown efficient an UPS, the of performance power exceedsthe thus and extension later a to due increases units connected the of consumptionpower the If needed. currently than powermore 20% with unit a buy torecommended generally is It UPS. the of capacity the calculating when included be should components additional foreseeable,the or planned are expansions system If 210 = 1.4 * in 150 VAvalue bycorrespondinggiven The is 150 W of nameplate on Rating example: Calculation used. be may 1.4 of factor a supplies,power switched-mode For constant.typical a by watt in rating the multiplying by calculated be must powerapparent the watts, in given is rating the If peripherals. and storage data external terminals,monitors, others, among together.added be simply include, Thesecan devices individual the of ratings the system, the of back the on label the in ratingVAon indicatedthe is If loads. non-linear forpowerapparent the calculating by nameplates,or the on specified (VA)amperes volt in output rated the on either based is Properdimensioning 24.3 Dimensioning Offline Technology

can be selected and adjusted via battery expansion sets so sets expansion battery via adjusted and selected be can

transtec = ttec in the Netherlands and Belgium and Netherlands the in ttec = transtec network. The shutdown software is installed on all other network components; network other all on installed is software shutdownnetwork. The the on availablecontroller.group data a UPS as the makes functions This and UPS the to serially connected is computer one created.that be means can This group a then UPS, a to connected aresystem operating network same the of etc.) workstations,gateways,routers,hubs, (servers, components network several If password.a by protected areetc. network the rebootingsettings, changing as Functionssuch rebooted.be tohave not does computer the even or software shutdown the that immediately.activated are workstation, means This every for made are which alterations,and edited be can users. messages network These pre-defined to time M network. the throughout workstation authorised every from managed be can they network, a in installed are UPSs several If network. the in workstation every fromsystem UPS a manage to possible is It Internet. the overeven possible restrictions) user-dependent (with access direct make options communication of execute automatically.variety toguaranteed A be should etc.)*.cmd, *.com, (*.exe,*.bat, scripts or instructions pre-defined UPS, the from comes message a If telephones. mobile GSM pager,email,or fax network, via itself UPS the of failure a or outage power a as situations critical for alarm the set to possible be should It Network-Wide Interfaces, 24.4 Alarm, Communication 25%. to up by lifetime battery the reducesCelsiusdegrees 5 temperatureby ambient the increasing that shownhave manufacturers Testsby out carried temperature. battery the by influencedgreatly is this years,but 5 and 3 between is life UPS. averageservice Their every of part essential an are batteries The It is also possible to define an exact plan to switch the UPS on or off.or on UPS the switch to plan exact an define to possible also is It UPS. the off Switching 6. and closuresystem proper a Performing 5. etc.), databasesavebackup, a (e.g.do commandsentered any Executing 4. applications, all of termination Correct 3. (Windows), files open all of state current the Storing 2. disk, hard the on contents memory cache the Storing 1. functions: following the includes software management UPS the by triggered shutdown A 24.5 Shutdown install. to easy is UPS the guide, user charge.the of tofree Thanks manufacturer relevant the of homepage the from downloaded be can systems.Updatesoperating common the all for software shutdown and management a with come USPsusually All computer.the to connection the provides which interface, USB or network a have also usually and interface RS232 a with equipped generally are devices UPS battery. the of status the judge correctly to possible it make will system. file UPS This the in medium storage energy the as battery the of importance the to due necessary, absolutely is file log battery separate a Keeping file. log a in savedare incidents All network. the in systems UPS the of all of test the initialise to workstation a at user a for possible be also should It components. network the of all of shutdown sequential controller.the from independent a specified example, ensures,for be This can members.group behaviour, as Their defined however,are components these essages, triggered by the UPS management software can be sent at a specific a at sent be can software management UPS the by triggered essages,

389

compendium