White Paper

Next-generation SONET for Cable MSOs: Technical Overview Next-generation SONET/SDH Bit rate ANSI Asynch SONET MSOs have several technology choices Name Tributary Transport available to prepare them for GbE-based 40 Gbps - - STS / OC-768 service delivery and the inevitable growth 10 Gbps - - STS / OC-192 in digital transport bandwidth. This 2.5 Gbps - - STS / OC-48 white paper provides an overview of 622 Mbps - - STS / OC-12 Next- 155 Mbps - - STS / OC-3 generation SONET digital transport 51 Mbps - - STS / OC-1 technology choices available to Cable 45 Mbps DS3 / T3 STS-1 SPE - Multiple System Operators (MSOs). 1.5 Mbps DS1 / T1 VT-1.5 - SONET standards were developed to 64 kbps DS0 - - provide robust, point-to-point, TDM- based signal transport across short, Table 1: SONET-based TDM payloads medium, long, and ultra long haul optical networks. SONET standards are hierarchal in nature and were pre- has overtaken TDM voice traffic on • Deliver more profitable services by dominantly designed for the transport today’s digital transport networks. leveraging today’s installed SONET of TDM-based voice signals. The table infrastructure below outlines the bit rates of hierarchal To satisfy the need to support traditional Next-generation SONET/SDH solu- SONET-based TDM payloads. TDM applications as well as emerging non-SONET data protocols such as GbE tions involve the implementation of sev- The line side optical bit rate of SONET and Fibre Channel, extensive “next-gen- eral key, standards-based technologies equipment has continued to expand to eration” standards development is and onto SONET/SDH-based transport meet growing voice transport require- has been taking place. The goals are to equipment. These standards include: ments. Over the years, the line side accomplish the following: • ITU-T G.7041 Generic Framing optical bit rate of SONET equipment Procedure (GFP) has grown from 155 Mbps to 622 Mbps • Provide SONET/SDH-compliant • ITU-T G.707/783 Virtual to 2.5 Gbps to 10 Gbps. 40 Gbps MAN and WAN transport of native, Concatenation (VCAT) SONET equipment is now emerging. non-SONET protocols such as GbE Traditionally, OC-192 (10 Gbps)-based and Fibre Channel • ITU-T G.7042 Link Capacity SONET transport was predominantly • Bring interface and protocol flexibility Adjustment Scheme (LCAS) used in long haul applications in an to SONET/SDH transport equipment • IEEE 802.17 Resilient Packet effort to pack as many TDM channels • Greatly improve service granularity, Ring (RPR) onto a single optical carrier. Due to the density, flexibility, and bandwidth reduction in the cost of OC-192 systems, use/efficiency for data-centric Generic Framing Procedure (GFP) this technology is now becoming cost- applications Generic Framing Procedure (GFP) is an effective for metropolitan area applica- • Drive intelligence into the SONET/ emerging standard defined by ITU-T tions as a means to expand metro area SDH transport layer G.7041. GFP is a generic Layer 1 data digital transport network bandwidth. • Continue to provide native delivery encapsulation mechanism designed to As SONET line rates have continued to of traditional TDM payloads accept and transport multiple native data grow, bandwidth and protocol flexibility protocols over metro and wide area digital of most SONET equipment has failed transport networks. It has been designed to keep pace. This is becoming increas- to stop the proliferation of SONET ingly important in metropolitan area mapping protocols (such as POS, x86 transport networks where many different non-SONET data protocols and services now need to be supported. In addition, it is now widely noted that data traffic

2 LAPS, and proprietary methods) for of the bandwidth in the client signal less idles and inter-frame gaps are not packet data services. GFP is deterministic contains needless idles (e.g. GbE inter- transported across the network. While with low overhead providing a robust, frame gap, preamble, start of frame GbE, Fibre Channel, ESCON, and high integrity, flexible mapping delimiter), these idles are FICON share common 8B/10B physical method for data protocols, such as transported in a transparent fashion layer coding, they do not share common GbE, into traditional digital transport resulting in the consumption of unneces- frame delineation and packet formats. network payloads. sary transport network bandwidth. As a result, a protocol specific physical Finally, transparent GFP requires only coding sublayer and MAC are required A common attribute of many widely minimal client signal awareness, thereby to extract and transport each protocol used, high-speed data protocols is their allowing a single hardware design to via frame-based GFP encapsulation. reliance on 8B/10B physical layer cod- transparently transport multiple proto- Therefore, different hardware designs are ing. 8B/10B coding involves the conver- cols based on 8B/10B coding. required to transport each of these sion of 8 bit bytes into 10 bit codewords protocols using frame-based GFP. to ensure high transition density and Frame-based GFP (GFP-F) DC balance (i.e. no long strings of 1’s Frame-based GFP involves the mapping Figure 1 provides a pictorial view of or 0’s) for clock recovery circuits on the of client data frames only into GFP both frame-based and transparent receiving end. frames. Inter-frame bytes of the client GFP encapsulation. Two types of GFP-based mapping have signal are not mapped into GFP frames. In conclusion, GFP-T and GFP-F offer currently been defined. These include Therefore, transport network bandwidth a differentiating set of capabilities and “transparent” (GFP-T) and “framed” is better utilized, especially on lightly permit a diverse range of data service (GFP-F). Each provides a unique set loaded client signals, because meaning- offerings. GFP-T is ideal for Storage of advantages over the other, depend- ing on the application. The current GFP Core Header GFP standard provides a clean map- GFP payload area comprising ping mechanism for emerging 64/65 B blocks (octet aligned)— all client encapsulated Transparent GFP Client input 8B/10B GFP-FCS Used for clients where the inter-frame gaps contains important client-specific data protocols, specifically GbE, Client PM Fibre Channel, ESCON, and information e.g. signaling information, flow control characters FICON. Additional proposals are SONET/SDH being considered to extend GFP-T 8B/10B 64B/65B 8B/10B for the support of DVB-ASI and Fast GFP map GFP 64B/65B decode decode demap demap encode (100bT). STS-x-nv to client

Transparent GFP (GFP-T) Fibre channel, ESCON

Transparent GFP involves the map- GFP Core Header ping of the complete client signal into GFP payload area comprising fixed length GFP frames, regardless of only client frames—not inter- Framed GFP Client input frame bytes (octet aligned) the information content of the signal. Used for packet-oriented clients— GFP-FCS no flow control or signaling characters Client PM It is intended for low latency trans- between packets mission of client data signals where the inter-frame gaps contain impor- SONET/SDH Remove Replace tant client specific data, such as sig- 8B/10B all inter- 8B/10B GFP map GFP all inter- naling information or flow control decode frame demap frame encode bytes STS-x-nv bytes to client characters (e.g. Fibre Channel and ESCON). If a significant amount Ethernet MAC frames, IP

Figure 1. Transparent GFP and Framed GFP

3 Area Network (SAN) applications while Supported characteristic GFP-F GFP-T GFP-F is better suited to Ethernet appli- Transparency to 8B10B—frame control codes no yes cations. The following table compares At full rate minimizes WAN bandwidth (removes IFG) yes no these GFP framing formats. Permits per-frame performance monitoring yes no Virtual Concatenation (VCAT) Permits sub-rate bandwidth (with frame granularity) yes no Virtual Concatenation (VCAT) is an Minimizes latency for delay-sensitive protocols no yes existing standard defined by ITU-T Long distance for Ethernet (local 802.3x PAUSE) yes no G.707/G.783 and ANSI T.105. VCAT Optional error correction for error-sensitive protocols no yes is not a new concept in principle. It was Permits sharing of transport pipe among multiple clients yes yes* originally designed to extend the utility of the SONET/SDH transport layer Table 2: Characteristics of GFP-F and GFP-T although it hasn’t been adopted to date as a mainstream networking technology. VCAT enables more efficient support of Service protocol Client rate Contiguous concatenation Virtual concatenation packet-based data services through the 100 Mbps STS-3c 65% STS-1-2v 98% “right” sizing of SONET/SDH channels ESCON 160 Mbps STS-12c 26% STS-1-4v 78% and bandwidth. Fibre Channel 850 Mbps STS-48c 34% STS-3c-6v 92% SONET/SDH transport structures, 1 Gbps STS-48c 40% STS-1-21v 92% including contiguous concatenation (i.e. STS-3c-7v 92% OC-3c/12c/48c/192c), are optimized for Fibre Channel 2 1.7 Gbps STS-48c 68% STS-3c-12v 92% traditional TDM transport applications and are rigid in nature. VCAT provides Table 3: Comparison of Contiguous Concatenation and Virtual Concatenation a highly granular, flexible, and efficient method of provisioning traditional GbE client signal (1,000 Mbps) can be ciency as well as much finer SONET/ SONET/SDH transport bandwidth for mapped to a VCAT group of 7 STS-3cs SDH bandwidth granularity. VCAT data-oriented service transport. Speci- (7 x 155.4Mbps = 1088 Mbps) to combined with GFP enables SONET/ fically, VCAT allows for the grouping achieve the same bandwidth efficiency of SDH transport elements to more effec- of non-consecutive SONET/SDH syn- 92 percent. Conversely, if traditional tively transport data-centric protocols, chronous payload envelopes (SPEs) to contiguous concatenation is used for such as GbE, in their native formats. create “virtual” concatenation bandwidth the same wire speed GbE client signal Link Capacity Adjustment groups. With VCAT, an arbitrary num- (1,000 Mbps), an OC-48c (2.488 Gbps) Scheme (LCAS) ber (X) of virtual containers (STS-1/3c would be required, thereby wasting SPEs) are grouped together with the approximately 60 percent of the provi- Link Capacity Adjustment Scheme (LCAS) is an emerging SONET/SDH combined payload (STS-n-Xv) used to sioned bandwidth. The following table standard and is defined in ITU-T match the required bandwidth. shows how VCAT can provide substan- G.7042. As we have indicated, VCAT tially improved bandwidth efficiency as As an example, a wire speed GbE client provides the ability to “right” size signal (1,000 Mbps) can be mapped to a compared to traditional SONET/SDH SONET/SDH channels and bandwidth VCAT group of 21 STS-1s (21 x 51.8 contiguous concatenation approaches resulting in more efficient support of Mbps = 1088 Mbps) resulting in a band- for various data protocols. packet-based data services. LCAS increases width efficiency of approximately 92 As displayed, VCAT results in improved the flexibility of VCAT by allowing the percent. Likewise, the same wire speed SONET/SDH transport capacity effi- dynamic reconfiguration of virtual concatenation groups.

4 Generic Framing Procedure • Efficient network resource utilization - Low GFP overhead Client signaling de-mapping - No over-provisioning with VCAT Payload multiplex • End-to-end service framing, PM and SP OAM, demarc Client signal VCAT buffering • Layer 2 independent (differential path delay) - Supports RPR, other Layer 2 protocols • Enables network consolidation - Converges next-generation services with existing infrastructure investment SDH/SONET DWDMOTIN

Virtual Concatenation • Efficient network resource utilization - “Right-sized” SONET/SDH bandwidth channels - Reduction in over-provisioning STS-n-Xv - End-to-end proficiency for packet services Payload inverse multiplex • TDM QoS, jitter performance Client-payload mapping - Deterministic QoS for all series • Newsoft protection schemes Interoperable global ITU-T standard— - LCAS (ITU-TG.7042) allow dynamic reconfiguration Mapping to SONET, SDH, and OTN of VCAT payloads - LCAS provides automatic signaling capability

Figure 2. Features and advantages of GFP, VCAT, and LCAS

LCAS eliminates the slow and inefficient Figure 2 exhibits the advantages of GFP, point-to-point connection and other provisioning process of classical VCAT, and LCAS. To employ GFP, nodes on the ring cannot claim unused SONET/ VCAT, and LCAS over an existing bandwidth. Resilient Packet Ring tech- SDH networks and offers a means to SONET/SDH network, only the termi- nology addresses these concerns and is incrementally enlarge or shrink the size nals at the end points of the connection aimed at improving SONET/SDH net- of a SONET/SDH data pipe without need be modified. This allows service work technology further. affecting the transported data. LCAS is providers to deploy GFP, VCAT, and used to add or remove members (STS-1s LCAS in a simple manner by installing Resilient Packet Ring (RPR) or STS-3cs) of a VCAT group resulting new tributary cards. Likewise, they can Ethernet is a low-cost, ubiquitous Layer in the “hitless” provisioning of more/less scale these implementations by adding 2 transport protocol that dominates bandwidth over a live SONET/SDH more tributary cards to any other termi- today’s LAN environments. The main VCAT circuit. LCAS uses a request/ nals in the network as required. drawback of Ethernet is that it is a acknowledge mechanism that allows On a final note, GFP, VCAT, and LCAS point-to-point, unprotected data trans- for the addition or removal of STS-1s/ have been added to the family of port protocol that lacks carrier-class STS-3cs to/from a VCAT bandwidth SONET standards to improve SONET attributes. A router or switch processes group. Finally, the LCAS protocol works protocol flexibility, bandwidth granu- each data packet at every hop in an unidirectionally, enabling service providers larity, and efficiency. Nonetheless, these Ethernet-based network. This attribute to provide asymmetric bandwidth. This improvements do not change the circuit- can be time consuming in larger Ethernet is particularly important for MSOs based approach of SONET as the point- networks, thereby resulting in jitter and requiring asymmetric transport band- to-point nature of each data connection latency performance that is typically width as seen in VoD deployments remains. Additionally, protection band- inadequate for real time services such as where content is distributed from VoD width continues to be reserved for every voice and video. servers to edge QAM devices.

5 Legacy SONET/SDH networks offer client access in the form of DS0, T1, DS3, OC-n, and in some cases, Ethernet. Today’s SONET/SDH networks offer exceptional jitter and latency perfor- Inner Outer Outer Inner ring data ring control mance and provide loss-less packet trans- ring data ring control port because packet data is transported from source to destination without the need for QoS decision-making in between. In addition, SONET/SDH networks reserve spare bandwidth capacity and provide 50 msec protection switching in the case of equipment faults or fiber cuts. Resilient Packet Ring (RPR) is an emerging Layer 2 protocol that is being Figure 3. Dual counter rotating rings of RPR technology defined by the IEEE 802.17 working group. RPR combines the low cost, Both physical rings are used to carry strip packets destined for them from the efficiency, and simplicity of Ethernet working traffic, thereby allowing band- ring thereby resulting in bandwidth con- with the high availability, reach, width traditionally set aside for SONET sumption only on the traversed segment. resiliency, and scalability of SONET protection to be utilized. This results in a Therefore, the unused spans on the ring transport technology. By combining doubling of available bandwidth capacity. remain idle and are available to other the advantages of both SONET and For example, an OC-12 (622 Mbps) stations and data streams. Spatial re-use Ethernet, RPR provides support for new, RPR ring provides 1.244 Gbps (2 x 622 is the term used to describe this property connectionless, packet-based services Mbps) of usable bandwidth. of RPR. while also providing traditional carrier- class features such as low latency and jit- RPR networks use a topology discovery All attached nodes in an RPR ring share ter, QoS, resiliency, and path restoration. algorithm such that all attached nodes the available transport bandwidth with- automatically learn the topology of the out the need for provisioning circuits. The result is the best of both worlds— network. In the topology map, each ele- The attached nodes automatically a resilient, packet-oriented, ring-based ment stores a primary and secondary negotiate for access to ring bandwidth solution that provides virtual mesh data path to every other element. Data is sent among themselves via a fairness control networking connectivity. via the optimal path only—typically the algorithm. Additionally, RPR provides RPR—Ring Operation shortest path unless a network failure express treatment of all marked packets, Resilient Packet Rings are ring-based condition exists. If a failure occurs, pack- thereby offering QoS for mission-critical networks that are optimized for connec- ets are automatically sent to the destina- traffic. Unlike Ethernet networks, RPR tionless data transport. As displayed in tion node via the secondary path within has a transit path architecture that allows Figure 3, RPR networks employ at least 50 msec. Additionally, destination nodes two counter-rotating physical rings used to connect all attached RPR network elements.

6 packets to quickly bypass intermediate As displayed, each RPR element contains • High Priority (Class A)—This nodes en-route to their destination. This two MACs to provide communication service provides Committed Inform- feature allows RPR networks to provide over the inner and outer RPR fiber rings. ation Rate (CIR) and guaranteed low latency and jitter performance for In a standard transaction, the host sys- bandwidth for high priority traffic time-sensitive services such as voice tem sends control information, including that is jitter and latency sensitive. and video. topology, fairness, protection, and OAM Traffic in this class is not subject to statistics, as well as data information to the RPR bandwidth-sharing algo- RPR Media Access Control (MAC)— the MAC control sublayer. The MAC rithm. Voice, video, and circuit emu- details and features control sublayer in turn sends RPR lation applications are ideal for this The IEEE 802.17 working group is cur- frames to the appropriate MAC for service class. rently defining the RPR MAC protocol. transmission over the physical layer. • Medium Priority (Class B)—This As with all Layer 2 protocols, RPR service is for CIR and other provi- defines a media access control (MAC) Packets received on either MAC are only sioned bandwidth applications requir- mechanism that defines the manner in routed to the host client if a destination ing less stringent (but still bounded) which available bandwidth on the physi- address match is encountered. All other jitter and latency requirements. cal media can be utilized by transmitting packets are placed back on the physical Access to ring bandwidth is guaran- stations. MAC protocols, including the ring media to transit toward other nodes teed for this service class as well. RPR MAC, also define how stations on the ring. This “transit” function Business data applications are ideal should react to congestion or collisions reduces packet delay significantly. for this service class. on the physical media. Finally, the RPR The RPR MAC offers four service classes • Low Priority (Class C)—This ser- MAC prioritizes and buffers packets, to the host client. These include: thereby providing regulated access to the vice provides best effort access to ring media. Figure 4 provides a high-level • Reserved—Unlike other services, the bandwidth. RPR nodes use a band- functional view of the RPR MAC. unused idle bandwidth for ‘reserved’ width-sharing algorithm to negotiate services is not available to other ser- for a fair share of the ring capacity for vices, thereby making this service traffic in this class. RPR ring band- class similar to TDM circuits. width is dynamically distributed for

Host client Interface to host system

Control Data

MAC control sublayer

RPR MAC RPR Layer 2 MAC defined by IEEE 802.17

Media access control Media access control

Outer ring Inner ring Physical media such as Physical Media Physical Media SONET and Ethernet

Figure 4. High-level RPR MAC functional view

7 RPR MAC traffic in this class. This service class MAC client interface is ideal for best effort, consumer traffic. H M L Receive data The RPR MAC has several prime func- tional blocks that collectively allow it to high medium low offer many advantages over legacy Ethernet and SONET/SDH networks. These functional blocks include the Drop logic Transit MAC client interface, drop logic, transit path, bandwidth manager, protection, topology, and physical layer as described Bandwidth manager in Figure 5. Physical IF The purpose of each of the functional blocks is as follows:

• MAC client interface—This inter- Protection Topology face allows the host to transmit and receive data based on the service class. The RPR MAC client interface pro- Functional blocks of RPR MAC vides the ability to back pressure the Figure 5. host system, thereby ensuring that transmitted packets have ring band- - Strips packets from the ring if bandwidth for each traffic class and width available. they are corrupted or their time enforces prenegotiated bandwidth • Drop logic—The RPR MAC to live has run out limits, as set by the RPR fairness inspects the destination address of all • Transit path—The transit path algorithm, for low priority data. The incoming packets to determine if the allows packets to quickly bypass inter- bandwidth manager also arbitrates packet should be stripped from the mediate RPR nodes as they travel to whether transiting packets from other ring or placed in the transit buffer their destination. It contains a small nodes or traffic from the local host for transmission to other nodes. high-priority and large low-priority will be placed on the physical media. The drop logic function performs buffer that can hold up to two data • Protection—RPR provides a 50 msec the following functions: packets. Packets are stored here protection mechanism for packet - Strips uni-cast packets from the while waiting for the local host to traffic. It uses packet steering as a ring and delivers them to the complete its current transmission. default with the option for packet MAC client interface if a match- The transit path is the key enabler wrapping. With packet steering, all ing destination address to that of of RPR’s ability to deliver reliable, stations are notified of a network the station is observed time-sensitive services. failure and transmitting stations - Copies multi-cast packets to the • Bandwidth manager—The RPR choose the ring that avoids the failed MAC client interface and places MAC tracks the bandwidth usage for span until they receive notification them in the transit path for trans- low priority packets and compares it that the failure has been cleared. mission to other nodes to bandwidth notification messages With packet wrapping, the nodes - Places uni-cast packets in the from downstream nodes. The system adjacent to the failure wrap traffic transit buffer if their destination decides how much bandwidth is to the other fiber ring. Other nodes address does not match that of available for low priority data and on the ring are notified and they the station whether local host client back pres- eventually steer their packets away from the failure until notification - Strips bandwidth notification sure should be applied. The band- that the failure has been cleared. packets and passes them to the width manager shapes and limits bandwidth manager

8 • Topology—The topology discovery on to the next node in the ring. RPR implementations algorithm of RPR allows attached Therefore, only one copy of a multi- Although RPR can be used to replace nodes to automatically learn the cast packet is sent around the ring. both SONET and Ethernet networks, it topology of the network. Therefore, • Carrier-class packet service perfor- is and will be used in combination with the host node can determine primary mance—With RPR, each node has both. For example, RPR is available on and secondary paths to all other two paths to every other node and present day SONET/SDH transport attached nodes and thereby choose packet steering is used to automati- equipment as well as on traditional the path a packet will take (typically cally protect against fiber or equip- IP/Ethernet routing equipment. the shortest path is chosen). The ment failures within 50msec. In topology map is updated at regular addition, RPR provides exceptional RPR and the physical layer intervals and after protection events. latency and jitter performance for The IEEE 802.17 working group is not This process allows nodes to be added mission-critical applications such as defining physical layers that are unique or removed from an RPR ring with- Voice over IP. to the RPR Layer 2 MAC. Instead, the out the requirement for a manual intent is to reference accepted and • QoS and fair access to ring band- provisioning process. proven physical layers for use by RPR. width—RPR provides multiple class- Physical reconciliation sublayers are used • Physical layer—RPR uses existing es of service for packets on or enter- to translate between a physical layer physical layer solutions, including ing the ring. New, differentiated interface and the common, standardized existing framers and optics. The packet services are easily accommo- RPR Layer 2 MAC interface. The stan- IEEE802.17 working group is devel- dated over a common network by dard currently defines such physical layer oping a reconciliation layer that will providing separate classes for interfaces for both SONET and support PoS and GFP encapsulated latency/jitter sensitive traffic, com- Ethernet physical layers. SONET and Ethernet physical layers. mitted information rate, and best effort traffic. The RPR fairness algo- For SONET/SDH physical interfaces, RPR performance benefits rithm ensures that each node is given two reconciliation sublayers have been Resilient Packet Ring technology offers a guaranteed amount of bandwidth, defined. The SONET/SDH reconcilia- multiple performance benefits, including: thereby preventing any single node tion sublayer (SRS) defines an HDLC- • Bandwidth efficiency—Because from being starved from access to like encapsulation for RPR frames that is RPR uses both working and protec- ring bandwidth. similar to PoS encapsulation. The GFP tion bandwidth for live packet traffic, • Ease of management—RPR offers reconciliation sublayer (GRS) defines an no bandwidth is wasted. Due to RPR’s plug-and-play simplicity, negating the encapsulation method using Generic transit path architecture and its need for manual provisioning when Framing Procedure. ability to use bandwidth only along nodes are added or removed from the For Ethernet, the GbE reconciliation the transversed segment while strip- ring. With automatic mechanisms for sublayer (GERS) defines an interface ping uni-cast packets at their destina- topology discovery and protection, between the common RPR Layer 2 tion, “spatial reuse” of bandwidth is RPR requires little or no operator MAC, and the GbE media indepen- achieved. This results in further management. dent interface (GMII). The 10 GbE bandwidth multiplication because • Scalability—RPR networks provide reconciliation sublayer (XGERS) unused portions of ring bandwidth high scalability by allowing over 64 defines interfaces between the RPR can simultaneously be used between nodes on a single ring. Nodes can be MAC and the 10 GbE media indepen- other nodes. RPR also provides a added or removed from a ring with dent interface (XGMII) and 10 GbE highly efficient mechanism for trans- topology discovery, bandwidth man- AUI interface (XAUI). mitting multicast traffic. Unlike agement, and protection algorithms meshed topologies where many packet automatically accounting for the copies are sent over multiple paths, change. Changes in bandwidth can be RPR nodes have the ability to receive accomplished quickly without com- a multicast packet as well as send it plex provisioning steps or truck rolls.

9 RPR on SONET/SDH equipment Ethernet Resilient Packet Ring On SONET/SDH optical network • Low cost • Scalability elements, RPR is implemented by • Simplicity • Reach Server integrating RPR-enabled Ethernet • Universality • Robustness interface cards and packet switching Optical capabilities to create an intelligent, Node distributed Layer 2 switch that uses server SONET/SDH bandwidth as a virtual backplane between end points. Figure 6 Optical is an illustration of this approach. Node Each RPR interface card is a highly SONET Distributed switch intelligent Layer 2 Ethernet switch. RPR cards come in multiple port Optical densities with various interfaces, such Optical Node as 10/100 BASE-T, 10/100 BASE-FX, Node and GbE. They support standard Server Ethernet protocols such as IEEE- 802.1Q and IEEE 802.1p. Server RPR-enabled SONET/SDH networks allow all, or a portion of, a ring’s total Figure 6. RPR Implementation on SONET/SDH SONET bandwidth to be provisioned as a ‘pool’ and dynamically distributed between multiple RPR interface cards RPR Ethernet cards provide Virtual RPR ring deployed throughout the ring. Each distributed Layer 2 switching intelligence Dynamic bandwidth pool dynamic ‘pool’ of provisioned band- Allocated RPR transport bandwidth for use by all RPR cards is dynamically distributed between all RPR-based SONET/SDH NEs width is referred to as a virtual RPR RPR Ethernet cards, thereby providing ring as displayed in Figure 7. A single virtual mesh connectivity network can support multiple, indepen- dent, virtual RPR rings if required. The key advantage of RPR-enabled SONET/SDH transport networks is the Virtual IPR ring ability to dynamically distribute band- Dynamic bandwidth pool width for packet traffic while continuing for use by all RPR cards to leverage the non-RPR portion of network bandwidth for TDM services. RPR BW Therefore, TDM services can be sup- ported in their traditional manner with TDM BW no loss of performance as opposed to utilizing packet-based circuit emulation methods. As a result, service providers can use their currently installed SONET/ Physical ring RPR-based SONET/SDH NEs 2 fibers—OC-12, OC-48 or OC-192 SDH infrastructure to deliver new STM-3, STM-16, or STM-64c packet-based, managed Ethernet services Traditional SONET TDM ring Dedicated bandwidth for circuit-based complete with SLAs, while continuing connections such as DS1/DS3/OC-n to deliver legacy TDM services with no loss of performance. Figure 7. TDM and RPR transport on common SONET/SDH network

10 IP router RPR on IP routing/Ethernet Physical interface switching equipment SONET/GbE/10GbE On many IP routing/Ethernet switching network elements, Ethernet is used as a client side interface and Packet over IP router IP router SONET (PoS) as a Layer 1 point-to- point WAN interface. Because RPR is supported on the SONET physical layer in a manner similar to PoS, it can be incorporated as a Layer 2 protocol into IP router RPR Layer 2 MAC implemented the WAN side of today’s IP routed/ throughout ring on WAN side interfaces Ethernet switched networks to provide highly resilient, virtual mesh connectivity. This is displayed in Figure 8. Additionally, RPR will facilitate the dou- Figure 8. RPR-enabled IP routed network bling of traditional PoS bandwidth. For example, an IP-routed OC-48c POS In addition to increased resiliency, RPR ring will have double the bandwidth employed on the WAN side of today’s capacity (from 2.5 Gbps to 5 Gbps) IP-routed/Ethernet-switched networks with RPR while continuing to provide will assist in significantly improving both 50 msec protection switching. Today, jitter and latency performance. This is Layer 2 and Layer 3 network topology due to RPR’s shortest hop, transit- information is exchanged between these based architecture. With many of network elements on a continuing basis. today’s IP/Ethernet or PoS physical ring- In the case of a network fault, Layer 2 based architectures, a router or switch or Layer 3 pro- fully processes each packet through the tocols, such as BGP and OSPF, must routing or switching engine in order to “re-converge” to route traffic around the make a forwarding decision to the next network fault. This process can, in many device. This results in higher packet loss, cases, take from several seconds to min- jitter, and unnecessary latency when utes to complete. Because RPR provides compared to RPR networks. 50 msec protection switching, when installed as a Layer 2 WAN interconnect The outstanding resiliency, bandwidth mechanism between IP routing equip- efficiency, and jitter/delay performance ment and/or Ethernet switches, network of RPR, combined with its QoS attrib- faults can be avoided without triggering utes, make this technology especially traditional Layer 2 and Layer 3 reconver- important for MSOs wishing to use their gence. With RPR, high layer protocols installed base of IP routing/Ethernet such BGP and OSPF are none the wiser switching equipment for the transport of that a fault has even occurred. mission-critical traffic such as VoIP. RPR is poised to become a dominant Layer 2 WAN interconnect mechanism for both core and aggregation IP-routed/Ethernet- switched devices within the MSO digital transport network.

11 Next-generation SONET Region Standard Description Status wrap-up Global G.7041 Defines GFP. Approved 2001 version + ‘02 As we have discussed, GFP, VCAT, addendum/corrigendum and LCAS are a collection of new and G.707 (sect 11) Defines VCAT. Also defines Approved contiguous concatenation. 2000 version + ‘02 existing standards designed to improve addendum/corrigendum SONET/SDH protocol flexibility, band- G.783 Detailed equipment specifications Approved width granularity, and efficiency. The (that include VCAT support requirements). 2000 version + table in Table 4 lists the standards that addendums G.7042 Defines LCAS. Refers to G.707 and G.783 Approved describe these protocols. for VCAT function. 2001 version + ‘02 addendum/corrigendum It should be noted that these protocols G.709/G.798 Definition + equipment specification do not change the point-to-point, for VCAT of ODUk. circuit-based nature of traditional IEEE 802.17 Defines Resilient Packet Ring MAC. First draft for ballot SONET/SDH networks. These proto- Q3 2002. Expected final standard cols are all Layer 1-specific. GFP, in mid 2003. both of its forms, is a Layer 1 frame ANSI T1.105 VCAT - Identical text to G.707/G.783 but Approved to SONET mapping protocol. It is not with SONET terminology. 2001 version GFP, LCAS - References G.7041/G.7042. designed to replace RPR, which defines ETSI EN300 417-9-1 Defines VCAT as per G.707/G.783. Approved a new Layer 2 MAC protocol. In fact, No recent updates to include GFP/LCAS. RPR can use GFP as a Layer 1 mapping protocol and the IEEE 802.17 working group is currently developing a reconcili- Table 4. Next-generation SONET standards table ation layer to provide this functionality. The GFP-based point-to-point data transport approach and RPR ring-based packet data approach enable different, Point-to-point RPR Comments complimentary services as described by Port aggregation x Point-to-point mapper: Just like any other TDM trib Table 5. RPR: Provides the ability to share bandwidth and Finally, it is important to note that aggregate to a single head-end port Layer 2 capable x Point-to-point mapper: Provide transparent transport multiple pre-standard implementations (logical customer of higher layer protocols. A front end Layer 2/L# device of RPR solutions have been on the mar- separation, CoS, etc.) would be required for customer separation. ket for some time. Specific examples of RPR: Integrated Ethernet LAN these include Nortel Networks OPTera Subrate x x Point-to-point mapper: Provided by WLAN bandwidth Packet Edge (OPE) and Cisco Systems management and flow control RPR: Provided by token bucket rate limiting and Dynamic Packet Transport (DPT) solu- flow control tions. However, the new RPR MAC Full rate over x Point-to-point mapper: Support for STS-24c, nxSTS-1 as being defined by the IEEE 802.17 SONET (n=21) and nxSTS-3c (n=7) Working Group fosters greater interoper- RPR: Provided by WAN load sharing and 2.4G RPR ability across both silicon and system Spatial Re-use x Point-to-point mapper: No bandwidth sharing implementations. The availability of RPR: Bandwidth shared between all stations on the WAN standard-based products will benefit Per port mappable x Point-to-point mapper: No bandwidth sharing end users by supporting a multi-vendor RPR: All Ethernet ports are shared environment. The IEEE 802.17 Fibre channel x Point-to-point mapper: USES GFP-T and a new trib Working Group was first formed in RPR: No support for PC December 2001 and the first draft stan- dard was published in August 2002. Table 5. GFP (point-to-point) and RPR comparison

12 Draft 2.0, code named “Darwin”, is the The OPTera Metro 3000 Multiservice most recent version of the standard and Platform series delivers increased net- the first to be circulated to a full working work efficiency and service offerings by: group ballot. The process will culminate • Delivering significant improvement with the publishing of the standard, in Capex/Opex costs which in the case of IEEE 802.17 and • Providing optimal utilization of RPR is expected later in 2003. existing assets Nortel Networks offers MSOs a family • Enhancing service delivery of market-leading next-generation • Enabling faster service deployment SONET platforms that sets a new eco- • Providing effective bandwidth nomic benchmark for cost reduction utilization while providing new differentiated data The Nortel Networks OPTera Metro services on existing networks. The 3000 Multiservice Platform series offers OPTera Metro 3000 Multiservice unique packaging options and architec- Platform series features a fully non- tural advantages that significantly reduce blocking switching architecture provid- the capital investment and lower the ing unmatched bandwidth management operational expenses associated with capabilities along with innovative service building and maintaining an optical modules enabling the industry's highest network. The OPTera Metro 3000 density service termination without Multiservice Platform series provides stacking multiple shelves. The OPTera forecast-tolerant and scalable optical Metro 3000 Multiservice Platform series networking solutions that enable service is equipped with Resilient Packet Ring providers to react to sudden changes technology that enables efficient band- in the marketplace without forklifting. width utilization with a full suite of native These next-generation SONET solu- rate client data interfaces. The OPTera tions cost-effectively serve applications Metro 3000 series is deployed by major ranging from customer premise to MSOs and enterprises and is consistently inter-office routes. recognized as the market leader.

13 List of acronyms HDT Host Digital Terminal PSTN Public Switched Telephone Network AM Amplitude Modulation HE Head-end QAM Quadrature Amplitude Modulation ATM Asynchronous Transfer Mode HFC Hybrid Fiber Coaxial QOS Quality of Service BLSR Bi-directional Line Switched Ring IEEE Institute of Electrical and RF Radio Frequency BW Bandwidth Electronic Engineers RPD Return Plant Demodulator CAP Control Access Protocol INM Integrated Network Management RPR Resilient Packet Ring CAPEX Capital Expenditure ISP Internet Service Provider SAN Storage Area Network CBR Constant Bit Rate IPPV Impulse Pay Per View SDH Synchronous Digital Hierarchy CMTS Termination System IP Internet Protocol SLA Service Level Agreement CO Central Office JIT Just in Time SNMP Simple Network Management CRM Customer Relations Management LAN Local Area Network Protocol DHUB Distribution Hub MAN Metropolitan Area Network SOHO Small Office, Home Office DOCSIS Data Over Cable System Mbps Mega bits per second SONET Synchronous Optical Network Interface Specification MPLS Multi Protocol Label Switching SRP Spatial Re-use Protocol DPT Dynamic Packet Transport MSO Multiple System Operator STB Set Top Box DVC Digital Video Compression NE Network Element STM Synchronous Transfer Mode DVB-ASI Digital Video Broadcast NEBS Network Equipment Building System STS Synchronous Transport Signal Asynchronous Serial Interface NSG Network Services Gateway TD Transparent Domain DWDM Dense Wavelength Division Ethernet to QAM Converter made TDI Transparent Domain Identifier Multiplexing by Harmonic Inc. TDM Time Division Multiplexing ETSI European Telecommunications OAM+P Operations, Administration, UNI User to Network Interface Standardization Institute Maintenance, and Provisioning VLAN Virtual Local Area Network FR Frame Relay OC Optical Carrier VOD Video on Demand GbE Gigabit Ethernet PC Personal Computer VoIP Voice Over Internet Protocol Gbps Gigabits per second PCM Pulse Code Modulation VPN Virtual Private Network GFP Generic Framing Procedure POS Packet over SONET WAN Wide Area Network PPV Pay Per View

In the United States: Nortel Networks 35 Davis Drive Research Triangle Park, NC 27709 USA In Canada: Nortel Networks Nortel Networks is an industry leader and innovator focused on transforming how the world 8200 Dixie Road, Suite 100 Brampton, Ontario L6T 5P6 communicates and exchanges information. The company is supplying its service provider and Canada enterprise customers with communications technology and infrastructure to enable value-added In Caribbean and Latin America: IP data, voice and multimedia services spanning Wireless Networks, Wireline Networks, Enterprise Nortel Networks Networks, and Optical Networks. As a global company, Nortel Networks does business in more than 1500 Concorde Terrace 150 countries. More information about Nortel Networks can be found on the Web at: Sunrise, FL 33323 USA www.nortelnetworks.com In Europe: For more information, contact your Nortel Networks representative, or Nortel Networks Maidenhead Office Park call 1-800-4 NORTEL or 1-800-466-7835 from anywhere in North America. Westacott Way *Nortel Networks, the Nortel Networks logo, and the globemark design are trademarks Maidenhead Berkshire SL6 3QH of Nortel Networks. All other trademarks are the property of their owners. UK Copyright © 2003 Nortel Networks. All rights reserved. Information in this document is subject to change In Asia: without notice. Nortel Networks assumes no responsibility for any errors that may appear in this document. Nortel Networks 6/F Cityplaza 4, GSA Schedule GS-35F-0140L Taikooshing, 1-888-GSA-NTEL 12 Taikoo Wan Road, NN104002-050703 Hong Kong