Download as pdf or txt
Download as pdf or txt
You are on page 1of 3

Advisory Report: Report Date: Analyst: Service: Market:

T erabit Optical & Data Networking Conference 2012: T ransport and Routing Roundup
May 11, 2012 Talbot, Rick Hot Topics , Service Provider Infrastructure Cloud Services , Data Center , Transport and Routing

Summary
Issue Layer123 and the Optical Internetw orking forum (OIF) hosted their second annual Terabit Optical and Data Netw orking conference in Cannes, France, April 16-19, 2011. This year, the organizers added an Executive Forum Europe, sponsored by The Optical Society (OSA) Corporate Associates. As last year, representatives of a variety of operators (mostly European) and transport vendors attended and made informative presentations on topics related to terabit optical netw orking and, new this year, cloud and data center netw orking. This years conference appeared to be more concrete than last years, a natural evolution considering the 400G trials (and Tbps laboratory demonstrations) that have taken place over the past year. Operators spoke more of how they are evaluating 100G transport, and those plans inevitably included their IP/optical integration strategies. W hereas presenters last year proposed super-channels last year as the method of delivering 400G and 1T transport, this year vendors presented how their upcoming products w ould employ these superchannels. In addition, vendors and operators agreed on QPSK and 16-QAM as the modulation formats for the new super-channels. How ever, operators, time and again, brought the discussions back to real-w orld planning issues revolving around 100G deployment strategy and how to optimize infrastructure for the rapid grow th of IP traffic. As operators focus on concrete planning, they are beginning to split their strategies for optimizing their netw orks. They are examining new netw ork architectures for meeting the challenge of exploding grow th in packet-based traffic. Simultaneously, they are considering more traditional methods of continuing to serve their legacy netw orks. In seeking to optimize their packet netw orks, they are seriously considering IPoDW DM for IP/optical integration, but IPoDW DM still faces serious control plane and netw ork management challenges in a multi-vendor environment.

Perspective
Current Perspective The presentations at Terabit Optical & Data Netw orking Conference 2012 addressed a number of themes, ranging from operator plans and evaluations of 100G transport to the latest developments in 400G and 1T transport. This year the topics included optical netw orking for cloud netw orking and for data center interconnections. Follow ing are discussions of some of the themes that w ill have the greatest impact on the optical netw ork infrastructure. Vendors Narrow Modulation Format Options for 400G Transport - Vendors appear to be in agreement on the tw o primary modulation formats for terrestrial 400G Transport QPSK and 16-QAM (both provided over tw o optical polarities). 16-QAM provides double the spectral efficiency of QPSK, but its strict optical signal-to-noise (OSNR) requirements limit it to a span of approximately 750 km before the signal requires regeneration. QPSK supports an amplifier span, w ithout Raman amplification and over standard (not ultra-low -loss) fiber, of 2,500 km to 3,000 km. Vendors suggest that operators can take advantage of the higher capacity of 16-QAM modulation in metro netw orks, but the vendors in this conference appeared to favor the simplicity of using the same techniques and equipment in both the metro and longhaul netw orks, implying that they w ould also choose the one modulation technique that w ould w ork in both environments. Some recent trials employed closely-packed sub-carriers in a super-channel that supported 4 b/s/Hz for QPSK (the same efficiency as 16-QAM w ithout the close sub-carrier packing), but those trials may not have observed the same conservative design requirements, such as a built-in loss margin, that may make such sub-carrier packing difficult to employ in the field. The primary significance of the selected modulation format for the future 400G transport systems is that, if these systems can achieve a spectral efficiency of 4 b/s/Hz, operators can continue to use the 100GHz ITU grid for their ROADM netw orks, and perhaps be able to upgrade existing systems. On the other hand, if these systems cannot achieve that spectral efficiency, the operators w ill likely need to deploy 400G systems on new fibers w ith gridless ROADMs. Operators Suggest Timing for 400G/1T Transport Demand - In one of its presentations, Deutsche Telekom asserted that it expected to need 400G transport in 2016 and 1T transport in 2022. BT claimed that it is now testing 100G transport, but has no immediate plans for 200G/400G (much less 1T). France Telecom - Orange mentioned plans to upgrade its core netw ork to 100G in 2014. TeliaSonera presented the trade-offs of using link aggregation of 10GbE connections betw een routers versus 100GbE connections, and concluded that it w ill sw itch to 100GbE w hen its cost has declined sufficiently. Colt Technology Services observed that there are a number of reasons to continue to deploy 40G connections rather than to rush right to 100G. Though Colt did not directly state the point, it implied that deploying 40G now incurs low er risk than deploying 100G. Finally, Level 3 indicated that it is considering the deployment of 100G transport in 2013. Taken together, these presentations indicate that the first 400G deployments are not just around the corner. Numerous operators are still just considering w hether to deploy 100G in the next year or tw o. Further, none of the operators raised the question of w hat concrete changes they w ould need to make in their netw orks to deploy 400G. Requirements Driving Networking Architecture on Two Tracks - Though most of the operators indicated that they provide both TDM and packet-based services, those w ho addressed netw ork optimization in presentations optimized their IP netw orks, not their TDM netw orks. None of these operators called their strategy cap and grow , but they are beginning to consider providing an optimized netw ork for their

high-grow th Internet traffic w hile employing more traditional means of serving their legacy demand. IPoDWDM Popular as IP/Optical Integration Method - As they presented their strategies for optimizing their IP netw orks, most operators expressed the desire for the efficiencies of IPoDW DM. These efficiencies minimize both OpEx and CapEx. IPoDW DM eliminates the need for a transponder at each router, w hich saves the CapEx for the space and cost of the transponder, and saves the OpEx of the pow er and management of the transponder. But Difficult to Manage - How ever, these operators observed that integrating IP netw ork and optical netw ork management is still a challenge, even in a single vendor environment, and a multi-vendor optical control plane solution has not yet been achieved. In fact, BT claimed that w avelengths generated by routers are, essentially, alien w avelengths in a ROADM netw ork. Cisco asserted that, because of the packet and optical netw ork management requirements for IP/optical integration, a single control plane for the router optical netw ork interface is not enough; common packet and TDM management are also required. Operators have yet to solve their IP/optical integration conundrum. They w ant to gain the benefits of IPoDW DM, but they concede that their router vendor is likely to be different than their optical netw ork vendor. The optical netw ork can carry w avelengths generated in the routers as alien w avelengths, but the effective interw orking of the IP and optical netw orks require a common control plane and possibly common packet and optical netw ork management. National Carriers Cool on Metro-Optimized 100G - All of the service providers w ho presented w ere national carriers rather than specialized metro carriers, such as those that focus on metro connections betw een data centers. Thus, even though these carriers appreciated the fact that some vendors have proposed cost-optimized metro 100G systems that use direct, rather than coherent, detection, they still preferred using a single coherent solution for their 100G systems in both metro and long-haul applications. This preference once again emphasizes the importance of controlling OpEx for carriers. They might save some CapEx on their metro 100G systems by employing the metro versions, but they do not choose to do so at the cost of the increased OpEx driven by managing tw o types of 100G systems. Additional Methods of Increasing Fiber Capacity - Operators and vendors, alike, have considered fiber bandw idth abundant, but fibers capacity is not infinite. The use of super-channels to provide 400G and 1T DW DM does not solve the ultimate transport capacity challenge; it merely uses fibers inherent capacity more efficiently. A DW DM system that employs 88 w avelengths carrying 100 Gbps each provides 8.8 Tbps of fiber capacity, and a system that employed 18 super-channels of 1 Tbps each (achieving a higher spectral efficiency via close packing of sub-carriers) w ould provide 18 Tbps of capacity. How ever, further raising of the super-channel capacity w ould simply decrease the number of channels that the fiber supports; the ultimate capacity of the fiber w ould have been reached, w ithout adding intermediate regenerators or using additional transmission bands (outside the C band) in the fiber. Some of the presentations suggested additional methods of increasing transport capacity w ithout lighting up new fibers or deploying a new fiber type (such as ultra-low -loss fiber). Operators could optimize the capacity of their netw orks through IP/optical integration. They could also broaden the optical bandw idth (up to 150 x 50 GHz channels) and improve a DW DM systems OSNR (enabling the system to employ a more spectrally efficient modulation format) w ith Raman amplification. In addition, the operator could employ an optical management system or control plane that routed sub-carriers through a netw ork of gridless ROADMs to maximize the total netw ork bandw idth utilization.

Recommended Actions
Vendor Actions Vendors should develop a deployment strategy for their future 400G systems that offers the benefits of QPSK and 16-QAM based superchannels w ithout striking the operators as complex. Vendors are currently proposing systems that provide either modulation format, based on system reach, but this flexibility may appear as a complexity to operators w ho simply w ant to know how the system w ill w ork. Vendors should focus their product roadmaps on the delivery of 100G DW DM rather than a future 400G. Operators are still determining how (or w hether) to deploy 100G transport in their netw orks. These operators consider the 100G technology as expensive, particularly compared to 10G technology that borders on commodity status. Too much vendor focus on 400G w ill appear to operators as ignoring their primary concern for optical netw ork evolution. Vendors need to develop an IPoDW DM-based IP/optical integration that w ill w ork in a multi-vendor netw ork. Operators clearly desire the efficiencies of IPoDW DM, and they appear to be focusing their optimization efforts on IP netw orks. How ever, because most operators employ optical and routing platforms from separate vendors, the control plane and management challenges of multi-vendor IP/Optical integration are hindering, if not thw arting, their plans. Vendors should determine w hether it is possible to achieve 4 b/s/Hz in a QPSK-based product, so that it w ould be capable of long-haul transport, yet use the ITU 100GHz grid. If not, they need to begin to consult w ith the Tier 1 operators to determine w hich changes in their netw ork required for the deployment of 400G DW DM w ould be most acceptable to them. User Actions Operators need to inform vendors of their preferences in the modulation format for 400G systems. If an operator w ill not consider deploying a cost-optimized metro 100G system due to the complexity of managing separate modulation formats for the tw o systems, it may not w ant to manage 16-QAM in the metro and QPSK, w ith half the capacity, in the long-haul. The vendors need to know w hether there is actually a market for cost-optimized metro 100G systems so that they can focus their development efforts on solutions that operators w ill use. Operators should consider the possibility of segregating their core into separate IP (or IP/MPLS) and traditional OTN netw orks. These carriers find themselves on the horns of a dilemma. They must continue to provide their legacy services (w hich may still be grow ing) as efficiently as possible, w hile preparing for the onslaught of IP traffic that could threaten to make their business unprofitable. A segregated IP netw ork w ould alleviate any organizational issues w hile allow ing it to deploy a targeted IP solution. Operators should encourage router vendors to develop a multi-layer control plane w ith separate transport vendors. Such an interoperable control plane is the minimum requirement for an IP/optical netw ork that can offer integration benefits. Some transport and router vendors have demonstrated the first versions of such connections, and operators can motivate a broader industry development of

multi-vendor control plane support by indicating their preference for such solutions. Operators w ho are now deploying 100G transport, and w ho see no end in traffic grow th, should begin to consider w hat changes in their netw ork infrastructure might be required to ease the deployment of 400G DW DM, and later 1T DW DM. The ultimate limitation of fiber capacity is not solved simply by increasing the super-channel capacity. Fiber capacity is increased by increasing the spectral efficiency of the system, likely by employing 16-QAM. How ever, a system that employs 16-QAM modulation w ill probably require an infrastructure change, such as closer amplifier spacing or additional regenerators.

All m aterials Copyright 1997-2012 Current Analysis, Inc. Reproduction prohibited without ex press written consent. Current Analysis logos are tradem ark s of Current Analysis, Inc. The inform ation and opinions contained herein have been based on inform ation obtained from sources believed to be reliable, but such accuracy cannot be guaranteed. All views and analysis ex pressed are the opinions of Current Analysis and all opinions ex pressed are subject to change without notice. Current Analysis does not m ak e any financial or legal recom m endations associated with any of its services, inform ation, or analysis and reserves the right to change its opinions, analysis, and recom m endations at any tim e based on new inform ation or revised analysis. Current Analysis, Inc. 21335 Signal Hill Plaza, Second Floor, Sterling, VA 20164 Tel: 877-787-8947 Fax : +1 (703) 404-9300 Current Analysis, Inc. 2 rue Troyon, 92316 Sevres Cedex , Paris, France Tel: +33 (1) 41 14 83 17 http://www.currentanalysis.com

You might also like