Professional Documents
Culture Documents
TPACK is focused on providing solutions that enable the transport of Ethernet data across telecommunication networks with Carrier Class quality. Ethernet over SONET/SDH and Ethernet over MPLS (including VPLS and PWE3) have proven to be the most popular methods to date.
However, with the inexorable growth in traffic, tighter integration of Ethernet with optical networks is desirable, in order to drive cost efficiencies and economy of scale. P-OTN promises to reduce capacity costs and network complexity, while improving scalability and flexibility. To better understand the driving forces and benefits of P-OTN, TPACK is providing this white paper which addresses the different approaches to P-OTN, including information on how TPACK can assist equipment vendors in capitalizing on the opportunities these technologies provide.
1) P-OTN (Packet Optical Transport Network) is the integration of packet networking technologies such as Ethernet and MPLS with optical network technologies such as WDM and ROADM. P-OTN provides the network infrastructure for NGN and supports multiple services including Carrier Ethernet. Copyright June 2008 TPACK A/S (www.tpack.com)
P-OTN:
From this strategic perspective, P-OTN encompasses the multitude of different system descriptions that have recently appeared, including PONP2, POTS3, CET4, and others less commonly used such as OPT5. These descriptions and possible distinctions are explained later in this paper. As network and system architects work to define flexible, cost-effective and future-proof designs, it is already clear that one size of system solution will not fit all network circumstances. Indeed, mainstream industry opinion holds that further evolution and incorporation of new standards from ITU-T, IEEE and IETF will be a defining characteristic of P-OTN over the coming years. In chasing such a moving target, the risk for both network and system implementers is twofold: Freezing a design too soon means early obsolescence, wasted effort and stranded investment, since interoperability with forthcoming solutions is compromised; Jumping forwards too far by adopting prestandardised approaches, or approaches that fail to be adopted by the wider market, carries exactly the same consequences. In both scenarios, carriers cannot achieve the economy of scale they need for profitability. Whilst for system suppliers, playing a waiting game is commercial suicide in today's cut-throat market regime - thus the tightrope must be walked between advancing too quickly, and being left behind. TPACK has built its business by helping systems houses to walk this line: allowing them to push their solutions forwards while building in the flexibility required for inevitable future changes. A partner white paper SOFTSILICON for Flexible Packet Transport explores TPACK's approach and the benefits that manufacturers can gain, ensuring that they stay afloat in this time of market turmoil and packet network transition.
2) 3) 4) 5)
PONP: Packet Optical Networking Platform POTS: Packet Optical Transport System CET: Carrier Ethernet Transport OPT: Optical Packet Transport Copyright June 2008 TPACK A/S (www.tpack.com)
P-OTN:
It is obvious that when designing practical network platforms, a variety of options for functional and technology integration exists and carriers are keen to leverage convergence to lower their TCO. Converging packet and optical is an attractive proposition, especially if it can be achieved with a flexible solution that can allow a phased migration from existing transport infrastructure based on SONET/SDH to a fully packet-oriented infrastructure. Two broad integration categories can be identified: Packet Optical Transport Platforms and Converged Packet Switch/Routers (colloquially, IP-over-WDM). With IP-over-WDM, a WDM interface or transponder is provided in a router (typically), which allows a WDM link between routers. The consequence of this approach is that packet switching decisions are made at layer 3 and requires this intelligence at each packet switching point. An alternative approach is to provide a layer 2 packet interface with a separate optical transport interface. Switching can be performed at layer 2, which is often desirable from a transport perspective. This is the basis for Packet Optical Transport Network (P-OTN). In this paper, we will concentrate on P-OTN, but for those interested in an overview of IP-over-WDM, please see the Cisco whitepaper Converge IP and DWDM Layers in the Core Network.
P-OTN:
Transport has historically referred to Layer 1 networks (eg. SONET/SDH or WDM technology), however the shift towards packet services means that Layer 2 now has an important role to play. Packet transport networks must enable carriers to drive the same economy of scale for NGN as they have done for legacy networks: Price/performance: achieving the lowest costper-bit transport; Service reach: providing the widest geographical footprint for customers; Multi-service: so that costs are shared across multiple lines of business; High availability: with low failure rate, fast protection and optional restoration schemes; High QoS: predictable latency, low errors and deterministic service delivery; Transparency: to handle any end-user or carrier's service unaltered;
P-OTN:
Strong security: to support any customer's data with confidence; SLAs: delivering on a carrier's promise for performance and availability. Some may question why, in a packet world, a transport network is still needed: since everything is going to be IP, all that's needed is routers, with intelligence centralised at Layer 3 and with integral IP-over-WDM ports for transport. However, on a per-bit carried capacity basis, the CapEx for routers is many times more expensive than alternatives and many carriers believe that their OpEx is also considerably higher. This drives a requirement for a separate, intelligent transport network at Layers 0/1/2 independent of Layer 3 switching/routing functionality. If L3 sophistication is not needed, then it should not be paid for.
P-OTN:
Phase 1 (2000-2003) - SONET/SDH + OTN: OTN served as the static WDM layer used to multiply fibre capacity. The principal SONET/SDH client was TDM, although Ethernet-over-SONET/SDH was beginning to increase in importance. Phase 2 (2003-2007) - OTN + Ethernet: SONET/SDH was still an important client, but GbE private lines developed to become a significant driver of deployment, together with some 10GbE. Lower capacity EoSONET/SDH private line services are also a significant new service on a flexible OTN layer. Phase 3 (2007 onwards) - P-OTN. With the introduction of L2 switching capability, packet and TDM services have become of equal importance. Ethernet is no longer bound to SONET/SDH as a client, but is carried directly by the OTH, after packet processing, aggregation, etc. ROADMs enable the OTN to be dynamically re-configured.
P-OTN:
Fig. 2: Evolution of OTN protocol stack and equipment architecture (simplified). Despite this illustration of a smooth evolution, and a logical progression towards convergence, in fact convergence drives carriers and vendors to different architectural solutions, depending on their starting points, their rate of investment, technological preferences, and their customers' requirements. For example, particular platforms or particular network nodes may not include L2 aggregation/ switching functions - whilst others may place TDM support as a secondary priority. Unlike the classical SONET/SDH networks, whose design tended to be proscribed by standards and homogeneous in nature, today's networks have evolved organically and display more variation. In fact, several system or platform architectures may be conceived, each optimised for a particular place in a carrier's network, or for a particular position in a system vendor's product portfolio. Despite the focus on convergence, network and system architects have much more choice in their specific deployment and evolution strategy.
P-OTN:
Fig. 3: Technology convergence options create P-OTP for P-OTN (picture adapted from Verizon)
Integrated SONET/SDH for existing transport interoperability, multiplexing/switching to support native low latency, high availability TDM services; Connection-oriented Ethernet (including MPLS PWE3, PBB/PBB-TE and T-MPLS/MPLS-TP) packet transport aggregation/switching and client interface support; Carrier-class OAM and management capabilities with optional ASON/GMPLS dynamic control plane. Many see P-OTP as a migration-enabling platform, and a common requirement is the need to support both TDM services and packet services with a strategy for smooth migration towards an all-packet infrastructure. This all-in-one approach could theoretically lead to a collapsed network architecture with fewer elements that could lead to simplified operations, maintenance, and management. Several of the available P-OTP single platform architectures build upon existing MSPPs or MSTPs with added OTN, ROADM and/or
P-OTN1:
Connection-Oriented Ethernet module support, while other platform architectures are based on universal fabrics capable of natively supporting any combination of TDM and packet traffic. However, it is also possible to deploy P-OTN using separate single-layer network elements, each optimised at the Optical and Packet switching layers. In theory this approach could help carriers leverage what they already have deployed in their networks, leading to cost savings via investment protection. In addition, this approach might also provide advantages in terms of further scalability and product maturity that today's single-element solutions might lack.
for packets. Firstly, CapEx forecasts for transport networks are not set to decline, but do show a shift in spending from SONET/SDH to P-OTN suggesting that P-OTN is earmarked to replace SONET/SDH. Secondly, emerging requirements on routers are emphasizing Layer 3 to Layer 7 support, such as Deep Packet Inspection and acceleration of applications. Analysts such as Infonetics Research track the trends in equipment investment, and the key evolutionary trends can be seen in Fig. 4. According to Infonetics, the market for optical network hardware will grow from $14bn to $15bn over the next 4 years with WDM growing at the expense of SONET/SDH. Note that the Packet Optical Transport System portion of WDM is growing from $1bn in 2007 to $2.5bn in 2011, explaining a large part of this growth.
Benefits of P-OTN
Evidence suggests that carriers are indeed interested in a separate, intelligent transport network optimised
P-OTN:
In contrast, analyst company Heavy Reading believes that the packet-optical transport systems market segment will grow from essentially $0 in 2007 to reach $2.8bn by the end of 2012, a CAGR of 110% from 2008-2012. Despite widespread agreement that SONET/SDH is not the long-term solution, even the most aggressive forecasts predict that SONET/SDH will remain with us for the next 10 to 15 years. What's more, it does not appear that SONET/SDH is being replaced by Carrier Ethernet Switches and Routers, but by P-OTN. . An integrated P-OTN approach offers many generic benefits to carriers: Supports multiple services: TDM (SDH, PDH), Ethernet, SAN, Video; Scales from 100% packet to 100% TDM, taking care of service migration; Scales from 2Mbit/s to 10/40Gbit/s in a single platform architecture; Any-port to any-port switching, for full node flexibility; Remote management eliminates expensive and slow manual intervention; Native timing and synchronisation support for user platforms/applications;
Card commonality with legacy MSPP, easing spares inventory; OTN-based transport protection and OAM, for carrier-class availability; Connection oriented Ethernet for carrier management; C/DWDM, Mux/Demux, (R)OADM for photonic networking; ASON/GMPLS for dynamic control and service provisioning; Carriers' drive towards a lower OpEx is supported due to native P-OTN interworking with other TDM and packet platforms. Integration of all L1 and L2 technologies in a single management system means that established operational procedures can be leveraged. Carriers' drive to lower CapEx is supported due to investment protection (eg. full reuse of installed NG-SONET/SDH base). In addition, if a carrier's future network evolution or demands are unclear, this approach avoids buying into a dead-end.
10
P-OTN:
11
P-OTN:
12
P-OTN:
Conclusion
Transport networks have historically delivered reliable, deterministic operation that can be interworked, if necessary, using systems from a wide variety of different vendors. Management systems tend to be well-organized and the network itself can be readily managed and maintained by good operational staff without requiring exceptional skills. The challenge in hand is to provide the same carrier benefits but within a packet-based infrastructure. P-OTN provides Carrier Ethernet/MPLS support, scalability as well as simplified packet networking, and hence deliver the lowest cost-per-packet-bit as well as the lowest cost-per-circuit-bit. It has a valuable role to play in the future NGN. The vision of an all-packet NGN supporting multiple IP-based services is not in danger. It has rather become more nuanced as the need for a broader set of supporting solutions has been identified. Transport is still important to carriers as a separate, operationally-driven domain of intelligence and expertise designed to assist and complement efficient IP-service delivery. In this regard, one can expect a bright future for P-OTPs in P-OTN, though one can also expect that evolution will continue in both architecture and naming. Watch this space!
13
P-OTN:
OTH Level 1 2 3
OPU client Payload capacity (Gbit/s) 2.488 320 9.995 277 40.150 519
OTU frame Optical line rate (Gbit/s) 2.666 057 10.709 225 43.018 414
14
P-OTN:
Fig. 8: OTN layers (source: ITU-T G.709/Y.1331). After this point, we leave the digital or electrical domain and enter the analogue or photonic domain related to WDM transmission with the Optical Channel (OCh) Optical Multiplex Section (OMS) and Optical Transport Section (OTS). As this whitepaper concentrates on digital issues, these OTN areas will not be discussed further. As Fig. 8 shows, client signals - Ethernet, SONET/SDH or other GFP-encapsulated clients - are mapped into an OTH payload and transported transparently and unaltered to its destination. The mapping procedure adds ODU headers that operate as the embedded management channel, performs error checking and correction, and so on. Fig. 8 illustrates the multiplexing of four ODU1 signals into an ODU2. The ODU1 signals including the Frame Alignment Overhead and an all-0s pattern in the OTUk overhead locations are adapted to the ODU2 clock via justification (asynchronous mapping). Four of these adapted ODU1 signals are byte interleaved into the OPU2 payload area, and their justification control and opportunity signals (JC, NJO) are frame interleaved into the OPU2 overhead area. ODU2 overhead is added after which the ODU2 is mapped into the OTU2. OTU2 Overhead and Frame Alignment Overhead are added to complete the signal for transport via an OTc signal.
P-OTN:
Link Fault Signalling (LFS) and proprietary use of IPG and preamble bits in the Ethernet header). Two non-standard techniques are described as solutions commonly used to resolve this. In clause 7.1, the solution is to use Constant Bit Rate for 10G (CBR10G) mapping of the signal into OPU2 and to
increase the transmission clock rate to accommodate the extra bits that need to be transmitted. This allows the full 10G LAN PHY frame to be transparently transmitted. However, since the clock rate is different to standard OTN and to avoid confusion, the various layers are designated as OPU2e, ODU2e and OTU2e respectively.
In clause 7.2, a similar technique is used for mapping to an OPU1 instead of an OPU2. It uses the 2.5G (CBR2G5) mapping described in G.709, which differs from CBR10G mapping in that it does not use Frame Stuffing (FS) bits. Overclocking is again used where the overclocked OTU1 (OTU1e) data rate is slightly less than that specified for OTU2e (11.0491 Gbps rather than 11.0957 Gbps). The various layers are designated as OPU1e, ODU1e and OTU1e. The final standard method of transmission described in clause 7.3 is designed to provide the same level of transparency for 10G Ethernet as is achieved using
GFP-T in SONET/SDH for Gigabit Ethernet (GE). However, since 10G Ethernet uses a 64B/66B block code rather than the 8B/10B code used by GE, idle characters need to be removed in order to fit into an OPU2. This is equivalent to asynchronous GFP-T transparent mapping in SONET/SDH. Since GFP-T is not suitable for 10G transmission, a modified version of GFP-F is used. Discussions are ongoing as to the evolution of these transmission methods, especially in the context of OTU3 and 40G transmission. There is a desire to make sure that standard methods will allow
16
P-OTN:
transparent transport of, for example, 4 10G Ethernet LAN PHY clients in an OTU3. The concern is that if this is not accommodated that proprietary solutions will again surface hampering interoperability.
At the time of writing, definition of other ODU payloads is in progress at ITU-T. The ODU0 frame is intended to directly support GbE services, whilst the ODU4 frame is intended to support 100GbE services. So, even though OTN is a relatively old standard, there is a lot of activity right now to improve OTN's ability to accommodate various Ethernet clients.
OTN references Document ITU-T G.872 ITU-T G.709 ITU-T G.sup43: ITU-T G.975.1: ITU-T G.870: Description Architecture for the Optical Transport Network (OTN) Interfaces for the Optical Transport Network (OTN) Transport of IEEE 10G Base-R in Optical Transport Networks (OTN) Forward error correction for high bit-rate DWDM submarine systems Terms and definitions for Optical Transport Networks (OTN)
See also ITU-Ts Study Group 15 (SG15) webpage for tutorials on OTN: http://www.itu.int/ITU-T/studygroups/com15/index.asp
OTN OAM
One of the significant OTN improvements on SONET/SDH is Tandem Connection Monitoring (TCM). In OTN 8 levels of monitoring are defined including Section Monitoring (SM) for point-to-point connections, Path Monitoring (PM) for end-to-end monitoring and 6 levels of TCM where the start and endpoints can be freely defined. TCM provides the possibility to nest and overlap monitoring layers to better fit real-life situations. In Fig. 9 an example is shown where a service is provided by a service provider across two other operator's networks. In this example, PM is used to monitor the end-to-end service. TCM1 is used to monitor transmission in the Service Provider Network. TCM2 and TCM3 are used for overlapped monitoring of Operator As network. TCM4 monitors Operator Bs network while TCM5 monitors the working and protected paths in Operator Bs network. Individual links between nodes can also be monitored using SM. As can be seen, the various layers can be cascaded or overlap depending on the need.
Significance of OTN
OTN is more than just an improvement to SONET/SDH. OTN also provides manageability to a growing part of carrier's networks, namely WDM. While WDM to date has traditionally been used as a high capacity link between nodes, the introduction of ROADMs has introduced the concept of switched WDM networks based on ring, meshed or partial meshed architectures. Managing such networks is critical to assure effective use of resources and customer satisfaction. With TCM, FEC and the ability to map various clients, both TDM and Packet, OTN provides manageability, ease of provisioning and greater efficiency. For example, by using FEC, it is possible to extend the distance of fibre transmission or reduce the power required to reach an equivalent distance, which could enable more DWDM wavelengths to be accommodated. SONET/SDH did include in-band FEC capabilities, but based on a less effective scheme than the Reed Solomon based methods used in OTN. One of the major advantages of FEC in relation to ROADM networks, it that it increases the number of nodes that can be crossed without suffering packet loss due to cumulative attenuation.
17
P-OTN:
Both of these chip solutions are open to adaptation and customization as well as acting as inspiration for customer defined chip solutions designed to fit a given system architecture. SMARTPACKTM P-OTN chip solutions are based on reprogrammable FPGA technology, which allows customers to adapt and customize the solution to meet specific requirements. Future unforeseen developments can also be accommodated quickly, which will be critical in P-OTN applications, since they are still under definition in many cases. System houses can minimize risks and shorten development cycles by considering a solution from TPACK. In a new and dynamic market such as this, the stakes are high and mistakes can be costly. The TPACK approach mitigates against these risks by offering a P-OTN solution that can be adapted quickly and easily. This provides system houses with a valuable tool in exploiting the opportunities that P-OTN provides. Contact TPACK for more information on the SMARTPACKTM P-OTN solution and on how TPACK can assist you in your P-OTN development projects.
18
P-OTN:
TPACK in brief
TPACK is one of the world's leading providers of embedded software solutions for packet transport. TPACK's solutions are based on Field Programmable Arrays (FPGAs) with pre-integrated and pre-tested driver, application and management software. Formed in 2001, TPACK has helped Multi-Service Provisioning Platform (MSPP) providers to rapidly develop Next Generation SONET/SDH (NG-SONET/SDH) linecard and microMSPP solutions for transport of Ethernet, IP/MPLS and VPLS packet data across the existing telecommunications network. Now TPACK is helping these and other equipment providers to address P-OTN, T-MPLS and PBT applications for reliable, carrier grade Ethernet transport using TPACK SMARTPACKTM P-OTN, PBT and T-MPLS SOFTSYSTEM and SOFTSILICON solutions. - SOFTSILICON can best be described as a Flexible ASSP: it is a standard chip solution similar to commercial off-the-shelf chips, but has the advantage that is based on programmable technology, namely FPGAs. With the latest advances in FPGA logic density and power consumption, it is now possible to offer a standard chip solution with performance similar to ASSPs. The advantage is that it takes less time to develop and upgrade these chip solutions in response to changing market conditions. SOFTSILICON thus provides a cost effective solution for Telecom Equipment Manufacturers who need the ability to react quickly to unforeseen market demands. - SOFTSYSTEM provides a tailored solution that can include SOFTSILICON products, but also software. SOFTSYSTEM supports adaptations and customizations based on specific customer requirements quickly and right-first time accelerating Time-to-Market with new system solutions. - SMARTPACKTM is TPACK's family of SOFTSILICON products and solutions targeting Carrier Ethernet
and Packet Transport applications. TPACK's SMARTPACK products include NG-SONET/SDH packet mappers and Ethernet/MPLS Carrier Packet Engines (which combine packet processing and traffic management in a single device). TPACK also provides Carrier Packet Mapper Engine products, which combine the functionality of NGSONET/SDH packet mappers and Carrier Packet Engines in a single FPGA device. This is made possible through the use of Stratix III Altera devices, which are based on 65-nm manufacturing processes. Stratix III FPGAs provide up to 50% power reduction compared to Stratix II devices allowing more compact, lower cost and lower power consuming MSPP and microMSPP solutions. - TPACK Carrier Packet Engines provide multiprotocol support of various packet protocols, such as PBB-TE/PBT, T-MPLS, MPLS, VPLS, PWE3 as well as standard 802.1ad VLAN/MAC switching. These protocols can be supported simultaneously with full OAM support and multiple parallel operations per packet without affecting throughput performance. - SONET/SDH line rates from 155 Mbps to 10 Gbps are supported. Together with NG-SONET/SDH standards (GFP-F, VCAT and LCAS), this allows packet data rates from 1.5 Mbps to 10 Gbps to be transported with high efficiency including full, transparent transport of 10 Gbps LAN Ethernet payloads. - TPACK's customers include Alcatel-Lucent, NEC, Tellabs, Xtera/Meriton Networks and Turin Networks. TPACK has over 130 design-wins with 8 of the top 10 optical transport equipment vendors who account for over 60% of the market. - TPACK is headquartered in Copenhagen, Denmark with a sales office in Palo Alto, CA, USA.
19
P-OTN:
Glossary
10GbE ANSI ASTN CapEx CET COE CWDM CY Demux DWDM E-LAN E-LINE EoSDH/SONET ETH ETSI Institute FEC FPGA GbE GFP GMPLS HDTV IEEE Engineers IETF IMS IP IPTV IP-VPN ITU-T L2 L3 LAN LCAS MPLS MPLS-TP MSPP MSTP Mux NASS NG-ADM NGN NG-SDH NPU 10Gbit/s Ethernet American National Standards Institute Automatic Switched Transport Network Capital Expenditure Carrier Ethernet Transport Connection-Oriented Ethernet Coarse WDM Calendar Year Demultiplexer Dense WDM Ethernet-LAN Ethernet-LINE Ethernet over SONET/SDH Ethernet European Telecommunications Standards Forward Error Correction Field Programmable Gate Array Gigabit Ethernet Generic Framing Procedure Generalized MPLS High Definition TV Institute of Electrical and Electronic Internet Engineering Task Force IP Multimedia Subsystem Internet Protocol IP Television IP Virtual Private Network International Telecommunication Union - Telecommunication Sector Layer 2 Layer 3 Local Area Network Link Capacity Adjustment Scheme Multi-Protocol Label Switching MPLS-Transport Profile Multi-Service Provisioning Platform Multi-Service Transport Platform Multiplexer Network Attachment Subsystem Next-generation Add-Drop Mux Next-Generation Network Next-Generation SDH Network Processing Unit OADM OAM ODUk OpEx OPT OPUk OTH OTM OTN OTUk PBB PBB-TE PBT PDH PON PONP POS P-OTN P-OTP POTS POTS PWE3 QoS RACS ROADM SAN SDH SLA SONET STM-n TCO TDM T-MPLS TSS VCAT VLAN VoIP VPLS VPN WDM XC Optical Add-Drop Mux Operations, Administration and Maintenance Optical Channel Data Unit, level k Operating Expenditure Optical Packet Transport Optical Channel Payload Unit, level k Optical Transport Hierarchy Optical Transport Module Optical Transport Network Optical Channel Transport Unit, level k Provider Backbone Bridge PBB - Traffic Engineering Provider Backbone Transport Plesiochronous Digital Hierarchy Passive Optical Network Packet Optical Networking Platform Packet over SONET/SDH Packet-OTN Packet-Optical Transport Platform Plain Old Telephony System Packet Optical Transport System Pseudo-Wire Emulation Edge-to-Edge Quality of Service Resource and Admission Control Subsystem Reconfigurable OADM Storage Area Network Synchronous Digital Hierarchy Service Level Agreement Synchronous Optical Network Synchronous Transport Module, level n Total Cost of Ownership Time Division Multiplexing Transport-MPLS Transport Service Switch Virtual Concatenation Virtual LAN Voice over IP Virtual Private LAN Service Virtual Private Network Wavelength Division Multiplexing Cross-connect
20