Principles Of Computer Networks And Communications

Chapter 9 1

Principles of Computer Networks and Communications Chapter Nine: Local area networks

Learning Objectives:

After reading this chapter, you should be able to:  Describe how different forms of LANs originated and how they evolved.  Differentiate LAN physical and logical topologies.  Identify LAN addressing issues and the role of MAC addresses  Describe the role of LAN segmentation and its impact on performance.  Compare and contrast Ethernet, Token Ring, and FDDI LAN models.  Describe the role of VLANs and LANE configurations in networking schemes.

Chapter Nine Outline:

9.1 Overview 9.2 LAN hardware and software 9.3 Ethernet: the once and future king 9.4 Improving traditional Ethernet 9.5 Token ring 9.6 LAN segmentation and interconnection 9.7 VLANs 9.8 Summary

Lecture Notes:

Overview  A local area network (LAN) o Relatively small span (office, small building, small campus) o More relevant to classify by link ownership: . LAN links are privately owned . WAN links are typically owned by public carriers o Two LAN classifications based on function: . Dedicated-server (server-centric)  Servers function only as servers  At least one server must be a file server  Better control and more secure than peer-to-peer  Majority of LANs in businesses are server-centric . Peer-to-peer  Each station is an equal (peer) of any other station  Any computer can access files on any other  Any computer can take on the server duties, although special functions can also be assigned

LAN hardware and software  LAN hardware and software are the concern of layers 1 (physical) and 2 (Data Link) of OSI and TCP/IP model architectures  These layers handle all the protocols and specifications needed to run the LAN  Higher layers become involved only in processing information and when LANs are interconnected o Layer 1 (the physical layer) . Transmission and receipt of bit streams via electricity or light . Physical specifications for device connection o Layer 2 (the data link layer) . Frame assembly and disassembly . Frame synchronization . Point-to-point flow and error control . Physical addressing . Medium access  Almost all LAN protocols are embedded in hardware and firmware on a network interface card (NIC) o NIC has ports to accommodate connectors that must be installed on each node of the LAN o Node: any device directly connected to the LAN medium or directly addressable on the LAN  Layer 2 addresses o Uniquely identify each addressable LAN device o Usually the same as the medium access control (MAC) device . Defined by the Institute of Electrical and Electronics Engineers (IEEE) . A physical address that is different for each NIC . Hard-coded by the manufacturer . Read into RAM on initialization . Every MAC address is unique, predetermined, and permanent . Uses flat addresses: identify individual machines, but contain no information as to the location of the machines or their relation to each other  Computers function as LAN servers and as user stations o Server computers differ from computer stations because they are faster and have more memory and disk space o Different LAN uses and demands require different servers o Print servers: use a technique called spooling where print jobs from LAN stations are put in a queue on the print server’s hard disk and sent to the appropriate network printer  Network operating system (NOS) o Mediates between the stations of the LAN, the LAN resources, and the processes being run o Is to a LAN what a computer operating system (OS) is to a computer

o Controls the remote hardware and software of the LAN to achieve the actions required of a LAN o Specialized NOSs (Microsoft Windows Server or Novell Netware) are installed separately from computers’ OSs . The complete NOS resides on the LAN server . Small segments of the NOS are installed on each station . Redirector: a NOS small segment that examines actions initiated on the local station  Directs local actions to the computer’s OS  Redirects actions needing a network resource to the LAN NOS  Channels incoming actions to the local OS for handling o NOS functions: . Incorporates protocols needed to run a LAN . A means for software on the LAN to use hardware of the LAN . Controls all server operations . Manages network disk access, file storage, and server memory . Manages file security . Provides network administrators with tools to handle the LAN

 Media are the physical links that tie LAN components together o LANs run on media types like coaxial and twisted pair cables, fiber-optic cables, and wireless o Each type is paired with appropriate connectors (example: wireless needs transmitting and receiving antennas) o Within the province of the network architecture physical layer

Ethernet: the once and future king  Ethernet has become the most widely installed LAN  Traditional Ethernet (the originally released Ethernet) o 10BASE5 . First commercial Ethernet . Was designed to run as a logical bus on a shared thick coax physical bus to which each station was attached . Name signified a 10-Mbps data rate, baseband signaling, and a 500-meter maximum segment span . One segment could have up to 100 nodes . Medium attachment unit (MAU)  Although thick coax provides a wide bandwidth and good resistance to EMIs, it is difficult to work with and has inflexible layout designs  To connect a station, the cable must be tapped, and a MAU must be connected to both the cable and the station.

 Original Ethernet protocol: Carrier Sense Multiple Access with Collision Detection (CSMA/CD) o Contention protocol (each station contends for access) o To avoid chaos, each station follows the layer 2 protocol that guides access to the bus o A station wanting to use the medium first must listen for activity on the bus: . If bus is in use, station must wait . If bus is idle, station can transmit immediately o To prevent a station from monopolizing a LAN, after a station transmits one frame, it must stop and repeat CSMA/CD procedure o Collision: two stations, hearing no activity, transmit at the same time and their frames are destroyed o Jamming signal: After “hearing” a collision, a station stops transmitting and sends out a high-voltage signal that any station recognizes as a collision notification o To prevent constant collisions, each station must wait a random amount of time (the backoff) before attempting to re-sense activity on the bus  An Ethernet frame has five fields, plus the preamble and the SFD o Preamble: 10101010 repeated seven times for frame synchronization o Start frame delimiter (SFD): 10101011 completes synchronization and alerts receiver of frame start o Destination address: MAC address of the recipient o Source address: MAC address of the sender o Type/length: . If its value is less than 1,158 it indicates the length of the data field . If greater than 1,158, it indicates what the network layer protocol is . Example: a value of 2,048 indicates an IP protocol o Data PDU: variable-length field containing the LLC PDU (All data form higher layers) o Frame check sequence (FCS): Uses CRC for error detection based on all fields in the frame except the preamble and SFD  The collision window concept o Ethernet does not use acknowledgements, so higher layers of architecture are needed o Collision window (slot time): the length of time it takes for a frame to travel from one end of the LAN to the other o Ethernet limits the maximum speed of the LAN (and thus the size of the collision window) and mandates a frame size of 64 bytes to prevent collisions caused by distance issues . For 10 Mbps Ethernet, the slot time is 512 bit times . 512 bit / 8 bits per byte = 64 bytes . Uses twice the collision window to ensure that the station is still transmitting (and listening for its own collisions) the entire time it takes for the frame to reach the destination and for a jamming signal to possibly return

o Two design elements ensure a station is still in the process of transmitting for at least twice the slot time: the maximum length (500 meters) of the bus and the minimum frame size (64 bytes) o Propagation speed: how fast a bit travels on the bus o Bit rate: determines how long it takes for a station to transmit a complete frame  Persistence o Persistence strategies: the ways in which stations can act after the carrier sensing step o 1-persistence: if the medium is idle, the station sends almost immediately: . Waits until after the interframe gap (IFG) passes between successive frames transmitted from a workstation  IFG provides time for the NIC to prepare a frame for transmission  For Ethernet, the IFG is 96 bit times . 1-persistence strategy has the highest incidence of collisions (whenever more than one station is sensing an idle line at the same time) o p-persistence: requires a station to transmit with a probability p after finding the medium idle . Reduces the chance of collisions . Each station randomly generates a send-decision based on p, thus reducing the likelihood that stations will transmit at the same time . The lower the p value, the less likely collisions are . If p = 1, p – persistence is 1-persistence o Non-persistence: on finding an idle medium, a station will wait a random amount of time and then sense the line again . Reduces likelihood of collisions . Increases delays in transmission, even if no other stations want to use the medium

Improving traditional Ethernet  Thinnet (cheapernets) o LANs using thin coax o 1985: IEEE released 802.3a, a thin coax version of Ethernet o Vs. Thicknets: LANs with thick coax o Advantages of thin coax: . Easier installation . MAU incorporated into the NIC rather than being a separate device . Lower cost than thick coax . Still able to resist EMI o Major disadvantage of thin coax: higher attenuation rates, so a shorter maximum segment span of the LAN o 10BASE2: thin coax version of the Ethernet . Segments can not exceed 185 meters . No more than 30 nodes allowed per segment

. Only four repeaters can be used (extending span to 925 meters)

 Thicknets o LANs using thick coax (diameter of a garden hose) o Use medium attachment units (MAUs) to connect to stations and cable o Provides wide bandwidth with good resistance to EMI o Max segment span: 500 meters o Total span limit: 2,500 meters o Difficult to work with and inflexible designs o Heavy cable with a large minimum bend radius (not good) o Vampire taps required to connect stations

 Star wiring o Physical star configuration o Still operates as a logical bus: central hub distributes signals from one station to all others . Active hubs: act as repeaters by regenerating signals that come to them . Passive hubs do not regenerate, simply distribute signals o 10BASE-T . Uses thinner, lighter, and more flexible unshielded twisted pair (UTP) . T stands for twisted pair . Stations are connected to the hubs with two pairs of UTP run in half duplex mode  One pair for transmission  Other pair for receipt and collision detection o Advantages . Improved reliability:  In physical star, a break only effects one station’s link . Improved management:  The hub is the central point from which each station can be traced via a simple network management protocol (SNMP) module installed in the hub . Improved maintenance:  Adding a station to a physical star only requires running UTP from the station to the hub o Disadvantages . Physical stars require much more cable than physical buses . The speed and span of the LAN is the same as the physical bus . The hub is a single point of failure . Moving to 10BASE-T from a coax LAN requires complete recabling . Collisions are still possible o 10BASE-FL . Same star configuration and data rate as 10BASE-T, but uses light- based hubs, NICs, and two multimode fiber cables instead of UTP . Provides immunity from EMI and greater span

. BUT: is a costly upgrade  Switches o Replaced the hub as the central device o Connects stations in pairs and will not connect a transmitting computer to a busy one o No contention for access: so LANs no longer operate as bus o Advantages of switches: . No collisions because there is no contention . Compatibility is maintained (MAC layer is left alone for backward compatibility) . Multiple pairs at a time can be connected by a switch . Simple to upgrade: simply remove the hub and plug cables into the switch o Disadvantages: . More expensive than hubs . Still the single point of failure  Fast Ethernet o Speed jump from 10 Mbps to 100 Mbps o Official designation is 100BASE-TX . In 1995 became the official IEEE standard, named 802.3u . To achieve 100 Mbps speed, bit duration was reduced . Two-stage encoding scheme:  First: 4B/5B block coding is applied  The result is encoded using multiline transmission – 3 level (MLT-3) o Similar to NRZ-I, but uses three signal levels: ( volts and 0 volts) instead of two o Start-of-bit transition for a 1-bit, none for a 0-bit . Advantages  Considerable speed boost  Backward compatible: 10- and 100-Mbps stations can run on the same LAN with autonegotiation (allows nodes to agree on a data rate)  Upgrade is simple if cat 5 UTP or STP is already installed: simply swap the NICs . Disadvantages  Rewiring is required if not cat 5 UTP or STP  NICs and switches must be replaced  Maximum segment length is 100 meters and total span is 250 meters o 100BASE-FX . Multimode fiber-optic version of 100BASE-TX . In the two-step encoding process, MLT-3 is replaced by NRZ-I . Immune to EMI . Increase in maximum span to 40 meters when running half duplex and 2 kilometers with full duplex

o 100BASE-T4 . Designed to run on cat 3 UTP . Requires four pairs of the lower-quality cable to achieve 100 Mbps:  Signals are split among the pairs to reduce the load on each  Three pairs are used to transmit (two duplex, one unidirectional)  Three pairs are used to receive (the same two duplex and the other unidirectional) . Maximum segment length of 100 meters . More efficient 8B/6T block encoding  Full duplex o 802.3x standard published by IEEE in 1997 o Can potentially double the speed of any half duplex Ethernet o Disadvantages: . Expensive: requires replacing the switches and NICs with full duplex versions . Full duplex only works over point-to-point connections, thus only star-wired LANs could be directly converted to full duplex  Gigabit Ethernet o Jump in speed to 1,000 Mbps o Backward compatibility: the frame and MAC layer were left alone o Minimum frame size increased from 64 bytes to 512 bytes (slot time 4,096 bit times / 8 bits per byte) o 802.3z IEEE standard for fiber-optic media . Adds an extension field that appends bits to the end of the frame as needed in order to bring the minimum to 512 bytes o 802.3ab IEEE standard for copper media o Two basic classifications of gigabit Ethernet: . 1000BASE-T  Runs on cat 5 UTP  Uses 4B/5B encoding  Has maximum span of 100 meters . 1000BASE-X  Uses 8B/10B encoding  Has three subdivisions: o 1000BASE-CX: a copper standard using twinax or quad cabling with a max span of 25 meters o 1000BASE-LX: a fiber-optic standard using 1,300-nm signals with max span of 300 to 550 m o 1000BASE-SX: a fiber-optic standard using 850-nm signals with a max span of 300 to 550 m o Principal demand for gigabit Ethernet is to support high data rates on backbones and in storage area networks o Has become a strong competitor to ATM

 10 gigabit Ethernet o 10GBASE-X o Latest approved standard development o 2002: released by IEEE as 802.3ae o Builds on gigabit Ethernet and leaves frame and MAC layer alone o Runs full duplex mode on fiber-optic media (7 versions): . 10GBASE-SR (short range) and 10GBASE-SW (short wavelength)  Use 850 nm multimode fiber (MMF)  Intended for distances up to 300 meters . 10GBASE-LR (long range) and 10GBASE-LW (long wavelength)  Specify 1,130 nm single-mode fiber (SMF)  Distances up to 10 kilometers . 10GBASE-ER (extended range) and 10GBASE-EW (extra long wavelength)  1,550 nm SMF  Distances up to 40 kilometers . 10GBASE-LX4  Uses wavelength division multiplexing (WDM) to carry signals on four wavelengths of light over one MMF or SMF 1,310 nm pair  Distances are up to 300 meters on MMF and 10 kilometers on SMF o In all versions, distances within ranges depend on cable type and quality o 10 gigabit Ethernet is cost effective as a high-speed infrastructure for both storage-area networks (SANs) and network-attached storage (NAS)

Token ring  1970s: Developed and commercialized by IBM o Marketed as a LAN that did not suffer from throughput degradation due to collisions and that had predictable and acceptable performance under all loading conditions  1982: IEEE published specifications for the token bus (802.4) o Principal use today: on manufacturing floors for equipment control in electrically noisy conditions  1983: IEEE standard 802.5  Although more expensive and complex than Ethernet, token rings had a large following in 1980s and 1990s in situations where reliable and predictable delivery of frames was most important and LAN loads were high o Provides deterministic, collision-free performance even when under load  Despite its significant installed base, token ring has a limited audience today.  Configuration and operation: o Physical star/logical ring configuration . Popularized by IBM

. Most common token ring configuration . Formed by connecting each station to a Multistation Access Unit (MAU) at the star center . Cabling is usually STP, although fiber is also possible o Local topology requires operation as a point-to-point link between each node and its two immediate neighbors (predecessor and successor nodes) o 802.5 does not specify physical topology: . Logical linkage forms a ring that can be implemented as a physical ring, bus, or star o Token: a small package that controls medium access . A station must have possession of a token to send a data frame . Only one token in circulation . Operationally, the token circulates around the ring, visiting each station in turn . When a station receives a token,  If it has no frame to transmit, it regenerates the token and sends it on  If it has data to transmit, it creates a data frame and sends that out . When a data frame is circulating, there is no token . As a data frame circulates around the ring, each station reads it:  If it is meant for another station, it is regenerated and sent back out  At the station it is destined for, the frame is marked as read and sent back out again, working its way around the ring until it arrives at the original station . Once the data frame response reaches the original station, that station removes the frame, creates a token, and send the token out to repeat the process . This process prevents any station from monopolizing the ring  Speed o The original token ring operated at 4 Mbps . May seem slow, but actually faster than Ethernet in heavy load operations . This is due to the absence of collisions and the token-sharing scheme o Later attempts to increase token ring speed went largely unnoticed due to the significant increase in popularity of Ethernet  Three frame types: token, data, and command o Frame fields: . Start frame delimiter (SFD):  Alerts the station to item arrival  The field contains particular code patterns so that frame type can be determined readily . Access control (AC): subdivided into priority (3 bits), reservation (3 bits), and token indicator (2 bits)

. Frame control (FC): indicates data frame or control frame, and type of control . End frame delimiter (EFD):  End of frame  Also used to indicate damaged frame and last-in-sequence frame . Frame status:  Used to indicate that a data frame has been read  Also terminates the frame . Source and destination address: MAC addresses that follow the same format as Ethernet (and all 802 MAC addresses) . Data PDU:  0 bits for token frames  Up to the maximum allowed by the particular implementation (based on ring speed)  Maximum total frames size is 18kb . Frame check sequence (FCS): like Ethernet, uses CRC

LAN segmentation and interconnection  LAN segmentation o The goal of segmentation is to reduce overall congestion by grouping stations together (segmenting them) according to traffic patterns . Segment creation is based on stations that most often need to communicate with either:  Each other  A common data source  Or, with a common resource o After segmentation, LAN traffic is largely isolated within each segment . This reduces overall traffic . LANs can later be interconnected to keep everyone in communication o Each segment must be a LAN in itself, with its own file server, hub/switch, and possibly other shared equipment as well o LAN segmentation increases overall performance . For example, suppose there was a 40-station 10-Mbps LAN . On average, each station will be operating at 250 Kbps (10 Mbps / 40) . If we reconfigure the LAN as two 20-station segments, then on average each will be operating at double the rate, 500 Kbps (10 Mbps / 20)  Bridges: o A bridge is a traffic monitor that sits between and connects to two LANs . Bridging reduces overall traffic by localizing segment traffic  Acts as a filter between the LANs to keep local traffic local and send crossing traffic across

o To filter the traffic, the bridge must know which addresses are on which sides: it keeps the addresses in a forwarding table o Bridge types are distinguished by how the bridge address tables are established: . Basic bridges:  Tables are manually loaded in a tedious process  Makes sense only in those that are unlikely to change (where stations are rarely added, or NICs rarely replaced)  Technical support needs to be readily available  Low in cost . Learning bridges:  Automatically create their own tables  Two simple versions: o When a frame shows up at the one port, the bridge puts the source address of the frame on the side of its table that corresponds with that port o Flooding: the bridge sends a special frame to the LANs on both sides, which is then repeated to every station. The response frames come back to the bridge, and the source addresses are entered into the table  When fully constructed, the bridge table will have a two- column list of all the addresses on the first side and on the second side  Subsequently, when a frame arrives, its destination address is compared to the side of the column corresponding to the LAN it came from. o If it finds a matching destination in the column it checks, it stays there o If not, it goes to the other column  Learning continues after this process, adapting to frames that aren’t in the table and handling LAN reconfiguration by periodic flooding o Transparent bridges . One bridge can connect more than two LANs . The bridge will have one port for the connection to each LAN and one column in its address table for each port . Operation is a simple extension of the two-port model . Transparent: the stations act as they normally do, and are unaware of the bridge o Translating bridges . Only bridge type that can connect LANs if layer 2 protocols don’t match . Operationally, quite complex

o Source routing transparent bridging . A type of translating bridging used to connect the two LAN types . Follows IEEE standard 802.1d . Has a transparent/learning side for Ethernet and a source routing side for token ring . The most straightforward interconnection solution  Redundancy o Redundancy allows continued operation in the face of some component failures . For bridged LANs, this means having more than one bridge o For the network to properly operate, there can only be one active path between two LANs . Loops  More than one active path  Can cause duplicate frames . Infinite looping  A frame is stuck in a loop and keeps following it around forever  Clogs up the network o Spanning tree method . Circumvents looping problems . Allows networks to achieve robustness by use of redundant bridges . Spanning tree:  Bridge ports are set up with one route from each LAN to each other LAN  The redundant routes remain unused until route failure calls for them  A tree structure is then overlaid on the network: o Each bridge has an ID, the one with the lowest ID is the root bridge o Designated port: the port on each bridge over which frames may flow o Blocking ports: all other ports that do not allow frame flow o Each bridge sends special frames called bridge protocol data units (BPDUs) out of each of its ports. . The root bridge calculates the “shortest path” from each bridge back to itself. . The ports that connect these paths are called root ports o The collection of allowed links will then have paths between every pair of LANs but no redundant paths, thus no loops o Ports on disallowed paths will not forward frames, and ports on allowed paths will forward them

o If a link or bridge fails, blocked ports can become designated ports by the same process as the initial setup  All of the work of setup and maintenance in a spanning tree is handled by the software and carried out automatically  Backbones o Efficient way to interconnect LANs o All interLAN links traverse the backbone o May be linked to bridges, based on routers, or they may even be LANs themselves . Whatever method, LAN stations connect to the backbone via their LAN hubs or switches and the backbone serves as a high-speed pathway interconnecting all of the LANs o Bridged backbone: . Each bridge has one port for connection to the backbone bus and another for connection to the LAN switch . A bridge will only forward frames from its LAN that are destined for a non-local LAN to the bus . A bridge will only forward from the bus frames destined for its LAN o Star-wired (Collapsed) backbone: . Each LAN switch is connected to a router that has tables of LAN addresses and will send frames from one LAN to another according to frame destination addresses . In this configuration, the backbone is considered to be shrunk into the router itself (hence the name) . Very popular configurations because routers…  Have powerful address-switching capabilities  Can be connected to external and internal links  Can be placed anywhere  Provide a single source for traffic management  Can incorporate remote monitor (RMON) devices and simple network management protocol (SNMP) software to permit easy traffic management . Drawback: if router fails, the entire backbone fails o Fiber Distributed Data Interface (FDDI) . Published as ANSI standard X3T9.5 and incorporated by ISO as a compatible version . Despite popularity in 1990s, has been superseded by higher-speed Ethernets . Runs at 100 Mbps . Stations can be up to 2 kilometers apart with single-mode fiber . Each station acts as a repeater (like a token ring) . Expensive due to its optical infrastructure . Reliability boosted by dual ring setup:

 Each ring operates simultaneously while traffic moves in opposite directions (counter-rotating)  If a station shuts down or a link on one ring crashes, the other ring picks up with virtually no time lost and ring operation is preserved  In effect, the ring is folding back on itself (wrapping) and becomes a single ring until the problem is fixed . Wrapping and reconfiguration are handled by the dual attachment concentrator (DAC) that attaches each station to the rings . For cost-relief, a copper wire standard of FDDI called CDDI was published by ANSI and ISO  CDDI is designed to run on either cat 5 UTP or type 1 STP  Distance suffers with copper wiring: limited to 100 meters  CDDI not suitable for MANs  Works well in backbone setups and is especially useful where cabling is already in place  CDDI is less complex and less costly

VLANs  Virtual LAN (VLAN) o The logical counterparts of physical LANs . VLANs accomplish via software what would otherwise require physically reconfiguring LANs o Grouped by station or switch characteristics, or frame protocols, without changing physical LAN membership or links o It doesn’t matter whether the stations are in the same LAN as long as there are physical connections among them o Four major benefits: . Security: messages and data transfers are not accessible to people who are not members of that VLAN . Traffic reduction: broadcast and multicast traffic can be restricted to the subsets of stations for which the traffic is relevant . Flexibility:  Easily set up and disbanded  Memberships are simple to add / remove  Stations can be part of multiple VLANs simultaneously . Cost savings: cost of a VLAN is minuscule in both money and time compared to the cost of physically moving stations and people o Oversizing a group or complex groupings may cause: . Congestion: unnecessary traffic on the connecting links can slow down all the stations using those links – even if they are not VLAN members . Network management difficulty: tracing problems can be tedious and time consuming, especially when the physical components are widespread

 VLAN membership: o Attribute based . Membership based on port numbers or station addresses . Switches are configured by creating list mappings/access lists that compromise a table of membership attributes (VLAN associations)  Stored in the switches  Switches use these to discern which ports belong to which VLANs, then they forward frames accordingly  Three means: o Mostly manual: . Network administrator enters the station assignment data . VLAN software eases this task: after the administrator enters defining characteristics, the software sets up the switch . Changes in membership must be manually entered o Partly manual: . Network administrator enters the initial assignments and defines the groups that the assignments fall into . If a member changes groups, switch reassignments are made automatically o Mostly automatic: . Administrator defines groups based on a characteristic . The members are automatically added or changed o Protocol based . Membership defined by frame characteristics  Decided on a frame-by-frame basis . Frame tagging  Most commonly used method for creating a protocol-based VLAN  IEEE standard 802.1q  Modifies the Ethernet frame to include tag information  The switches use this information to transfer frames to their corresponding VLANs  Advantages: o Easy way for one station to belong to more than one VLAN at the same time o Added level of security: each frame carries its own VLAN identification

 Disadvantages: o When several tagged VLANs are overlaid on the same physical internetwork, management and troubleshooting are greater than for port-switched VLANs o Also: additional processing needed to reconfigure the Ethernet frame if problems occur o Devices processing the frame must be compliant with 802.1q, otherwise tagged frames will be rejected  LAN Emulation (LANE) o Pseudo-LAN type o Most often applied to an asynchronous transfer mode (ATM) network that can transfer traffic between Ethernet or token ring LANs when it functions in LANE mode . As such, the ATM would be serving as a backbone . ATM LANEs are most commonly employed to simplify integration of Ethernet LANs with ATM networks . In either case, the process maps LAN MAC addresses to ATM cells and ATM cell addresses to LAN frames

End-Of-Chapter Questions:

Short answer

1. How are LANs classified? answer: LANs used to be classified by span (relatively small spanned), but link ownership is more relevant today. LAN links are privately owned, versus WANs’ public carrier ownership. However, function classification help distinguish between LAN types. Two basic LAN function classifications are dedicated-server (server-centric) and peer-to- peer. There are also distinctions made on the basis of protocols contained within the network operating system, physical and logical topologies, and media.

2. What are the layer 2 functions involved with LANs? answer: Layer 2 handles all of the protocols and specifications needed to run the LAN; it involves frame assembly and disassembly, frame synchronization, point-to-point flow and error control, physical addressing, and medium access.

3. How is the uniqueness of MAC addresses assured? answer: MAC addresses are 48 bits long. The first 24 bits (Organizationally Unique Identifier, or OUI) are assigned by the IEEE and are exclusive to each manufacturer. The last 24 bits (serial numbers) are assigned by each manufacturer and are unique for each NIC it makes. Because each NIC MAC address begins with an OUI, MAC addresses from different manufacturers will be unique even if they happen to have the same serial number.

4. Describe CSMA/CD. answer: Carrier Sense Multiple Access with Collision Detection (CSMA/CD) is a layer 2 Ethernet protocol that guides access to the bus. A station wanting to use the medium must first listen for activity on the bus. The station can only transmit if the bus is idle, otherwise the station must wait. Since the stations must contend for use of the bus in a first come, first served manner, CSMA/CD is called a contention protocol. To prevent monopolization of the LAN, CSMA/CD requires a station to stop and repeat the CSMA/CD procedure after every frame it transmits.

5. How do 10BASE5 and 10BASE2 differ? answer: 10BASE5 was the first commercial Ethernet, designed to run as a logical bus on a shared thick coax physical bus to which each station was attached. It has a 500-meter maximum segment span and after using its four repeaters, a maximum overall span of 2,500 m. In order to connect a station, the cable in 10BASE5 is tapped with a vampire tap and a medium attachment unit (MAU) is connected to the cable and the station. 10BASE2, however, uses thin coax cables and has a 185-meter maximum segment span due to a higher attenuation rate than 10BASE5. No more than 30 nodes are allowed per segment; thus with the four repeaters, the maximum overall span is still only 925 m.

6. How does the operation of Ethernet change when a hub is replaced by a switch? answer: When a hub is replaced by a switch, the LAN no longer operates as a bus because the stations no longer contend for medium access. When using a switch, collisions are eliminated and multiple pairs of computers can potentially be connected at the same time. Throughput potential is increased, but actual performance is still limited by access to servers.

7. How does LAN segmentation improve performance? answer: After a LAN is appropriately segmented, traffic is largely isolated within each segment. This reduces overall traffic. Segmentation also increases performance by increasing speeds within segments. Segments can achieve higher throughput when they are separate than they could if they were in their original LAN form.

8. Explain the operation of a learning bridge.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall Chapter 9 20 answer: Bridges must know which addresses are on both of its sides in order to filter traffic. To do so, bridges keep the addresses in a forwarding table, to use as a reference when a frame reaches them. Learning bridges can create tables on their own, automatically, thus removing the tedious process of manual table loading necessary in basic bridges. Learning bridges construct tables in two ways. In one version, the bridge records the frame source in the appropriate side of its table upon the frame’s arrival. The other way involves the flooding of the LANs on either side with a special frame. The response frames that come back to the bridge are then entered accordingly into the table.

9. How can VLAN memberships be defined? answer: VLAN memberships can be defined by attribute (switch port number, station MAC address, layer 3 IP address), or by frame protocol.

10. What is LANE? answer: When an asynchronous transfer mode (ATM) network is operating in a LAN Emulation (LANE) mode, it can transfer traffic between Ethernet or token ring LANs. ATM LANEs are most commonly employed to simplify integrating Ethernet LANs with ATM networks. The process involves mapping LAN MAC addresses to ATM cells and ATM cell addresses to LAN frames.

Fill-in

1. In a peer-to-peer LAN, each station is an equal of any other station.

2. The OSI and TCP/IP layers of primary concern to LANs are layer 1 (physical layer) and layer 2 (data link layer).

3. Almost all LAN protocols are embedded in hardware and firmware on the network interface card (NIC) .

4. The medium access control (MAC) address is the physical address of the NIC.

5. The NOS mediates between the stations of the LAN and the LAN resources.

6. The simplest device for connecting two independent LANs is a bridge .

7. To connect LANs with different protocols, a translating bridge can be used.

8. Fiber distributed data interface (FDDI) is a fiber-optic token passing dual ring.

9. A vitual LAN (VLAN) accomplishes in software what otherwise would require physically reconfiguring LANs.

10. Four major benefits of VLANs are security , traffic reduction , flexibility and cost savings .

Multiple-choice

1. In a dedicated server LAN a. a server also can function as a station b. a print server is required c. at least one server must be a file server d. stations can take on server duties e. all of the above answer: C

2. A network interface card a. has ports to accommodate connectors for the medium being used b. plugs into the system board c. may take the form of a PC card d. must be installed in every node of a LAN e. all of the above answer: E

3. MAC addresses are a. flat b. hierarchical c. determined by the network administrator d. software based e. geographically based answer: A

4. With Microsoft Windows Server and Novell Netware a. small segments are installed on each station b. the complete NOS is installed on the file server c. the station segment incorporates a redirector d. network disk access, file storage, and server memory are managed e. all of the above answer: E

5. The standard Ethernet frame a. has a maximum of 1,500 bytes b. prevents collisions by using tags c. depends on p-persistence d. prevents one station from monopolizing the LAN e. none of the above answer: D

6. A 1-persistence strategy a. means that a station can transmit at any time b. requires a station to wait a random amount of time after sensing an idle medium c. requires a station to transmit immediately after sensing an idle medium d. is a special case of p-persistence where 1 – p = 1 e. none of the above answer: C

7. Switch-based Ethernets a. eliminate collisions b. can connect more than one pair of stations at a time c. are a simple, inexpensive upgrade from hub-based Ethernets d. are the configuration used by Ethernets beyond 10BASE2 e. all of the above answer: E

8. With a token ring LAN a. collisions are impossible b. star-wiring is typical c. stations contend for access d. performance drops linearly with load e. both a and b answer: A

9. In a collapsed backbone a. the backbone is contained in a router b. individual LANs connect via bridges to the backbone c. there is a single source of failure d. no more than six LANs can be connected e. both a and c answer: E

10. AVLAN a. is a permanent reconfiguration of LAN membership b. is rarely used in business applications c. can cause congestion if not sized properly d. may be difficult to manage e. both c and d answer: E

True or false

1. The vast majority of business LANs are server-centric. T:

2. Ethernet LANs require NICs, but token ring LANs do not. F: A NIC with token ring protocols is required for a device to be connected to the token ring LAN

3. LAN stations are computers, but LAN servers are not. F: Both LAN servers and LAN stations are computers

4. A NOS is to the LAN as an OS is to the computer. T:

5. File servers cannot act as print servers. F: File servers can act as print servers

6. Windows XP and Mac OS incorporate the basic functions of a NOS. T:

7. Star-wiring is the required configuration for switch-based Ethernets. T:

8. Bus-wired LANs use more cable than star-wired LANs. F: Star-wired LANs use more cable than bus-wired LANs.

9. Each LAN segment must be a complete, independent LAN. T:

10. Attribute-based VLAN membership is based on port numbers or station addresses. T:

Chapter 9 – The MOSI Case - Part 1:

LAN goals (gleaned from the scenario):  To facilitate document transfer and sharing among the staff and management  To pave the way for a database application and electronic data processing  To handle transaction volume without bogging down  To be capable of easy upgrade and expansion when warranted by additional growth of the business

What questions would you ask of the managers, employees of MOSI or, other parties? Who will be connected to the LAN?

What applications will each LAN user need?

Are there any anticipated high-bandwidth user requirements?

Where will the users be physically located when they connect [e.g., Will they all be located in the same room? Same building? Same site? Will they be connecting from remote locations (e.g., home)? etc.]?

When will the information need to be accessed?

Which entities outside of MOSI need to access the information?

What computer resources are already available [e.g., operating system(s), storage, processing speed, NICs, etc.]?

What network resources are already available [e.g., cabling (cat 3, cat 5, fiber-optic) in the building and between buildings, wall jacks, servers, etc.]?

What networking expertise does the company have [e.g., Is anyone trained in managing a LAN? Who has experience in LAN security? etc.]?

Table that shows the results of the investigation: Notes:  Since there is not an existing token-ring network in place (and since the network does not involve real-time communications between embedded microcontrollers that might better be served by ARCnet), the proposed network will be designed using the industry standard Ethernet protocol.  Coax Ethernets are no longer a feasible network configuration  An assumption is that the data volume and access requirements will not rise to the level where a Storage Area Network might be required (at least initially).  Similarly, a 10GBASE-X alternative will not be considered here  If new wiring is required, it will be cat 5 or higher.  Before extensive wiring / re-wiring is considered, wireless networking alternatives should also be considered; however, these will not be included in the table comparison here.

Configuration Media Max Segment Speed NIC Cost Switch Cost [per w/s] 10BASE-T Cat 3 UTP 100 m 10 Mbps $171 $1302 100BASE-TX Cat 5 UTP 100 m 100 Mbps $173 $1304 1000BASE-T Cat 5 UTP 100 m 1000 Mbps $255 $2506 10/100BASE-FX Multimode 400 m 10/100 $1357 $5008 fiber Mbps Notes:  Switches are 24 port

1 Costs obtained from the Dell Computers online site, 2/24/2008 http://accessories.us.dell.com/sna/category.aspx? k=1000BASET&_nks=true&c=us&category_id=5584&cs=19&l=en&s=dhs&x=2&y=4 2 Costs obtained from Best Buy online site for Linksys 24 port 10/100 Ethernet Switch http://www.bestbuy.com/site/olspage.jsp? skuId=8308805&type=product&id=1173577950280&ref=06&loc=01&ci_src=14110944&ci_sku=830880 5 3 Costs obtained from the Dell Computers online site, 2/24/2008 http://accessories.us.dell.com/sna/category.aspx? k=1000BASET&_nks=true&c=us&category_id=5584&cs=19&l=en&s=dhs&x=2&y=4 4 Costs obtained from Best Buy online site for Linksys 24 port 10/100 Ethernet Switch, 2/24/2008 http://www.bestbuy.com/site/olspage.jsp? skuId=8308805&type=product&id=1173577950280&ref=06&loc=01&ci_src=14110944&ci_sku=830880 5 5 Costs obtained from the Dell Computers online site, 2/24/2008 http://accessories.us.dell.com/sna/category.aspx? k=1000BASET&_nks=true&c=us&category_id=5584&cs=19&l=en&s=dhs&x=2&y=4 6 Costs obtained from Best Buy online site for D-Link 24 port 10/100/1000 Ethernet Switch, 2/24/2008 http://www.bestbuy.com/site/olspage.jsp? skuId=8728442&st=Ethernet+Switch&lp=4&type=product&cp=1&id=1201306975378 7 Cost obtained for the AFP2X00 NIC from the Fiber Optic Cable Shop, 2/24/2008 http://fiberopticcables.stores.yahoo.net/etnetincar10.html Note: this module also includes a 10/100BASE-T/TX copper port 8 Costs obtained from L-Com online site for SWTC-FGSW2624SF switch, 2/24/2008 http://www.l-com.com/item.aspx?id=9573

Cabling cost comparison Media Cable Cost – 10 ft Cable Cost – 50 ft Cat 5 UTP $12 $40 Multimode fiber $33 $42

Notes:  Cat 5 UTP prices obtained from Radio Shack online site, 2/24/2008  Multimode fiber prices obtained from MWave.com online site, 2/24/2008

Chapter 9 – The MOSI Case – Part 2:

Corporate changes being considered (gleaned from the scenario):  Hiring additional personnel  Creating a marketing department and a legal department  Reconfiguring the scheduling and accounting operations as departments  Adding many more fee-for-service care providers

What questions would you ask of the managers, other employees of MOSI, and other parties? [All of the questions from Part 1 still apply.]

Would it make more sense to expand the current LAN to cover all personnel, or to have interconnected department LANs, and What is the business case for either decision, either way? At some point, LAN growth results in a loss of efficiency as more users contend for information on the network. In this scenario, it seems likely that users from different departments will need to access information and resources from other departments. Segmenting the network might relieve some of the contention issue by distributing some of the network load. However, the company should consider running separate LANs for some departments while enabling interconnection for information across the LANs; bridging might accomplish this. The business case involves considering what the company can afford to do, and the level of performance the company is willing to accept. At some point, the level of network use might require using a backbone (that is more efficient than bridging, particularly when employees are located on multiple floors or, buildings at a single site). Whether the separate LAN switches are bridged to the network backbone, or connected directly to a router (collapsed backbone) must be part of the considerations. As the complexity of the network expands, support personnel must have the expertise to design, configure and maintain the network. Cost of training employees must be weighed against the same service provided by outside personnel (i.e., outsourcing). Risk must be considered. As the company splits into separate interconnected LANs, the possibility of single points of failure must be examined and considered. Some