最新40G与100G供应商调查报告

Who Makes What: 40- & 100-Gbit/s Systems Who Makes What: 40- & 100-Gbit/s Systems 40/100G transmission embraces a range of technologies and has potentially wide applications – from transoceanic networks, through the metro, and on into equipment backplanes – so its long-term impact is likely to be considerable and widespread. For this Who Makes What, however, the main interest is taken to be telecom network transport applications, which means essentially: 40G Sonet/SDH (OC768/STM256) and the higher-rate ITU-T OTNs (OTU3/OTU4) Longer-range versions of 40 and 100G Ethernet. For more on 40/100G in equipment practice, especially ATCA, etc., see 40- & 100-Gbit/s Technology & Components. The list of operators and carriers now moving to, or with, some 40G implementations now contains some big names (see Table 1 for some recent examples reported by Light Reading), although not everyone is convinced by the 40G argument (see Page 4: Sticky Questions). In contrast, 100G is still largely experimental or trialing, but implementations are vaguely beginning, as witnessed by financial exchange NYSE Euronext and Ciena Corp. (Nasdaq: CIEN) announcing, in March 2009, plans to implement 100G networks to support NYSE Euronext’s new data centers in the greater New York and London metropolitan areas during 2010. Table 1: Recent 40G Network Implementations Reported by Light Reading Carrier/operator Location/date Implementation AboveNet USa/May 2009 40G metro service in 15 US markets Bell Canada Canada/September 2008 40G optical backbone network China Telecom China/August 2008 40G transmission network China Unicom China/December 2008 40G WDM transport network Deutsche Telekom Germany/July 2008 40G DWDM core network KPN Belgium/September 2008 40G upgrade to optical backbone network Lightower Fiber Networks USA/ June 2009 40G bandwidth service for carriers and large enterprises Mediacom Communications USA/December 2008 40G upgrade to regional network supporting triple-play services Neos UK/March 2008 40G network for delivery of bandwidth-on-demand for UK businesses Rascom Russia/July 2008 40G upgrade to long-haul optical network RoEduNet Educational Network Romania/November 2008 40G-ready network linking national educational and research facilities Southern Cross Cables USA/June 2008 40G upgrade to 10G terrestrial feeder to transpacific cable network Sprint Nextel USA/November 2008 40G transatlantic transmission trial SURF Telecoms UK/December 2008 40G upgrade to regional optical data network Telefónica Spain/November 2008 40G transmission network Telus Canada/December 2008 40G network upgrade TransTeleCom Russia/May 2008 40G commercial transport network connecting Moscow and St. Petersburg Triton Telecom Caribbean/October 2008 40G system linking Florida, Puerto Rico, Dominican Republic and Jamaica Virgin Media UK/May 2008 Lights 40G path Source: Light Reading, 2009 40G is also hitting the data center. In late 2008, for example, Mellanox Technologies and Dell claimed the world’s first demonstration of 40G InfiniBand interconnect technology for blade servers by using Mellanox’s recently launched InfiniBand ConnectX Adapter. The combination of hi-tech R&D, many smaller specialist companies, market evolution, and a global recession must make 40/100G one of the few current bright spots for M&A types, judging by the number of mergers, acquisitions, and similar announcements made over the last year or so. Examples are: ​ Aegis Lightwave Inc. acquired CardinalPoint Optics (April 2008 – optical channel monitors) and AOFR (March 2009 – fused fiber couplers) ​ Avanex and Bookham merged (April 2009 – optical components, modules and subsystems) to form Oclaro Inc. ​ EXFO Electro-Optical Engineering Inc. (Nasdaq: EXFO; Toronto: EXF) acquired PicoSolve (February 2009 – optical sampling oscilloscopes for 40G and 100G R&D) ​ Finisar Corp. (Nasdaq: FNSR) merged with Optium (August 2008) ​ GigOptix Inc. (OTC: GGOX) acquired Helix Semiconductors (January 2008 – optical physical-media-dependent ICs) and merged with Lumera (March 2008 – modulator technology) ​ Opnext Inc. (Nasdaq: OPXT) acquired StrataLight Communications (January 2009 – 40/100G products and subsystems) ​ Thorlabs Inc. acquired the assets of Covega (March 2009 – indium-phosphide and lithium-niobate components and modules) from owners Gemfire Corp. . Previously Gemfire and Covega had merged (February 2008). This list emphasizes the point, expanded on later, that optical devices and modules are crucial to 40/100G, and are where a lot of current product development is taking place. We have tried to make the listing as complete as possible in the time available for its compilation, but this is where you, Dear Reader, can help with any companies that have been missed. Environment & Technology Basic 40G transmission systems have been around for a while in ultra-long-haul and long-haul applications, but 40G is now spreading through the core to other parts of the network, particularly for metro ROADMs, as 40G becomes more generally deployed (see Table 1). Market research company Ovum Ltd. has estimated that by late 2008, more than 30 network operators worldwide had spent over $250 million since 2005 in deploying 40G. And the future looks potentially bright. According to market research company Infonetics Research Inc. , the global market for 40G and 100G optical network equipment should grow at a compound annual growth rate (CAGR) of around 46 percent from 2009 to 2011 to reach US$5.1 billion. Of course, the current economic slowdown is clouding the issue, and Infonetics stresses that, for optical networks generally, much depends on whether service providers do follow through during the second half of 2009 with their indicated spending. Vendors are naturally responding. An early commercial product exemplifying of the trend to end-to-end 40G is Huawei Technologies Co. Ltd. ’s E2E 40G Internet Protocol (IP) and optical transport solution, announced in February at the MPLS & Ethernet World Congress 2009 in Paris. This supports IP-based ultra-broadband services by combining the company’s NetEngine5000E high-end router and 40G DWDM transport system. Essentially, the move to 40G now and 100G later (but the sequence is disputed – see Page 4: Sticky Questions) rests on the standard telecom bandwidth-growth story. Bandwidth demands on the core are increasing because of the growth of IP traffic, video, and so on. Video traffic, in particular, has emerged as a major driver for more bandwidth – Cisco has estimated that video will represent half of consumer IP traffic by 2012 (and total IP traffic will be six times greater then than in 2007) – and 40G is seen by its adherents as a key way of handling router interconnection in this environment by aggregating multiple existing 10G links onto a smaller number of 40G ones. Another very important driver is fiber exhaust, both on metro and long-haul routes, which makes it important to squeeze more traffic onto existing fiber. Although it is possible to stack multiple 10G streams onto fiber to achieve a 4x10G equivalent to 40G capacity, for example, 40G supports point out that this is not efficient in terms of router port usage – a 4x10G interface on a router port is less efficient than a single 40G one. Although such a single 40G interface is more expensive than a single 10G one, the cost factor (which has improved considerably over the last couple of years or so) is likely to approach about 2.5 times that of a single 10G interface. So the industry is hoping that 40/100G will score twice with operators: It will cope with the traffic/bandwidth explosion and will save money by both reducing port counts (and associated operational expenditures) and extending the life of existing fiber assets. 100G 100G is still under development, but first standards are scheduled to appear end-2009 and during 2010 (see below). There have been various demonstrations and trials, and research initiatives established, as Table 2 indicates. Table 2: Recent 100G Demonstrations, Trials & Initiatives Company/organization Location/date Activity AT&T1, NEC Corporation of America and Corning Inc. USA/May 2008 Transmission of data at 114G over each of 320 separate optical channels on a single 580km optically amplified link Banverket Sweden/April 2009 Live field trial, running 10G/40G/100G simultaneously on existing fiber network between Sundsvall and Stockholm Deutsche Telekom and Ericsson Germany/March 2009 100G R&D field trial on existing optical platform as part of European 100Gbit/s Carrier-Grade Ethernet Transport Technologies Project Ciena USA/August 2008 Demonstration of single wavelength transmission of a 100G data stream, through 80 km of fiber with Caltech Georgia Institute of Technology USA/March 2009 Establishes the Georgia Tech 100G Optical Networking Consortium with 10 companies to perform multidisciplinary research in all aspects of 100G transmission Global Access and Infinera Japan/January 2009 Complete Japan's first 100G Ethernet demonstration between Tokyo and Osaka Neos Networks UK/March 2009 Trial of 100G DWDM optical system between Manchester and London Verizon USA/September 2008 Moves 100G trials program to next stage with over 1000km runout on Richardson, Texas, network Source: Light Reading, 2009 Particularly in the context of a 100G optical transmission system, it’s important to be clear on the difference between the external (client) interface and the system’s internal (line) interface. Although a system may present a 100G client interface, the fiber line transmission could be handled by, say, 10 optical wavelengths of 10G each (10x10G format), or by multiple 40G optical wavelengths. The goal, however, is to match both the client and line rates, so that a single 100G optical signal goes onto a single DWDM wavelength – which is why Ciena trumpeted its Caltech demonstration in Table 2, as it said that this was the first true, single-wavelength transmission of a 100G data stream through 80 km of fiber. Standards From the point of view of more traditional telecom protocols, 40G has been standardized for some time – it’s 40G Synchronous Optical Network (Sonet)/SDH OC-768/STM-256, which can be transported in turn by the International Telecommunication Union, Standardization Sector (ITU-T) ’s G.709 digital-wrapper technology (Optical Transport Network – OTN) 43-Gbit/s OTU3s. Work has been under way for a couple of years or so on the next level up for OTN, which is very logically OTU4 and, equally logically, is intended to be available to transport 100G Ethernet. The ITU’s Study Group 15 recently (April 2009) approved an amendment to the 2003 second version of Recommendation G.709/Y.1331 that "specifies 100Gbit/s ODU4/OTU4, support of gigabit Ethernet services via ODU0, ODU2e, ODU3 and ODU4, multi-lane OTU3 and OTU4 and the Lower Order and Higher Order ODU concept to align with the 'one-stage multiplexing' specification described in clause 9.2 of Recommendation G.872." Which presumably means in standards-speak that all technical bases are covered. Ethernet, and especially 100G Ethernet, is the real focus of interest because of its relentless rise as a universal Layer 2 network technology to underpin IP. Since December 2007, the Institute of Electrical and Electronics Engineers Inc. (IEEE) ’s P802.3ba Task Force has been working to define both 40 and 100G Ethernet standards, with a target date of 2010 for completion. Because Ethernet is now used over a range of scales, from meters to tens of kilometers, the eventual standards will embrace different Physical Layers, those most relevant to telecom network application being: ​ 10km range on singlemode fiber – both 40 and 100G ​ 40km range on singlemode fiber – 100G only However, these are not going to provide a heavyweight 100G telecom transport technology for the metro or core, and are really oriented towards campus-style networks and (along with the shorter-range versions) connectivity for client interfaces. Not only is the range too short, but there are major issues of spectral efficiency. Instead, the key importance of 100G Ethernet to telecom networks is as a standard 100-Gbit/s client interface to network equipment. In summary, 802.3ba aims to: ​ Support full-duplex operation only ​ Preserve the 802.3/Ethernet frame format utilizing the 802.3 MAC ​ Preserve the minimum and maximum frame sizes of the current 802.3 standard ​ Support a BER better than or equal to 10E-12 at the MAC/PLS service interface ​ Provide appropriate support for OTN ​ Support MAC data rates of 40 and 100 Gbit/s For a short overview of further aspects of 40/100G Ethernet standards, see 40- & 100-Gbit/s Technology & Components. Overall, this activity means that the 40/100G client interface is Ethernet, and the WAN interface is either OTU3/OTU4 or Ethernet for shorter spans. As always, interoperability issues, gap filling, and other matters will loom large in the commercialization of 40/100G Ethernet, and various industry initiatives have already sprung up. For example, the short-lived Road to 100G Alliance (later merged with the Ethernet Alliance) formed a technical committee in June 2008 “to identify gaps in technology and standards that could impact the rollout of 100G OTN and Ethernet networking platforms,” as the Alliance states. Multi-source agreements (MSAs) have long been a feature of the telecom equipment industry, as they commit groups of suppliers to supporting certain standardized module or device form factors, interfaces, and characteristics – and 40/100G is no exception. In March 2009, for example, Finisar, Opnext, and Sumitomo Electric Industries Ltd. / ExceLight Communications Inc. formed the CFP MSA, whose aim is to define a hot-pluggable optical transceiver form factor to enable 40 and 100G applications, including 40GbE, 100GbE, OC-768/STM-256, and OTU3 protocols, multimode and singlemode fiber optics, and various link distances. The earlier XLMD-MSA, formed by Eudyna Devices Inc. , Mitsubishi Electric Corp. (Tokyo: 6503), NEC Electronics Corp. , Oki Electric Industry Co. Ltd. , Opnext Inc. (Nasdaq: OPXT), and Sumitomo Electric Industries in 2007, had the more limited aims of establishing compatible sources of 40G optical transmitter and receiver devices embedded into 40G optical transceiver modules. In between, in May 2008, the Optical Internetworking Forum (OIF) , concerned that device and module vendors in particular were beginning to scramble in different, incompatible directions over 100G, formed a consensus to use DP-QPSK modulation with a coherent receiver in 100G long-distance DWDM transmission, and later launched two more initiatives to look at photonic integration and the use of Forward Error Correction (FEC) in such systems. Vendor & Product Categories The basic approach of this Who Makes What is to use three broad product categories, interpreted as follows: ​ Systems. The high-level network/transport systems or entities sold to network operators as the end product. This usually means at a minimum an optical transport product or family incorporating 40/100G WDM/ROADM capabilities, but may include switches/routers with 40G (currently) interfaces. ​ Devices / Modules / Subsystems. This forms a very broad category, covering the wide range of hardware elements used in end-product 40/100G systems. To make this category manageable, it is biased towards hardware devices and assemblies that provide 40/100G-specific functional capabilities. For example, modulators/demodulators, transponders, multiplexers/demultiplexers, network processors, and SerDes chipsets – but not lasers themselves, photodiodes, optical couplers, wavelength switching modules, or fiber, for example. Some items operate purely in the electrical domain, such as SerDes chipsets, while others are optical or part optical, such as laser modulators. It’s worth stressing that, in the optical-transmission world, "module" often means an assembly in a standard form factor, such as 300-pin or SFF. ​ Test & Measurement. This is a self-evident category, but biased towards the optical aspects of 40/100G. Table 3 lists vendors against these broad categories for 40G and 100G applications. A main idea is to use further Tables to show vendors’ general product types or specific products to clarify the distinctions within the categories, to give more information, and to distinguish between 40 and 100G product activities when necessary. Table 3: 40/100G Vendors Vendor 40G systems 40G devices / modules / subsystems 40G T&M 100G systems 100G devices / modules / subsystems 100G T&M ADVA Optical Networking Yes Aegis Lightwave Yes Yes Yes Yes Agilent Technologies Yes Yes Alcatel-Lucent Yes Altera Yes Yes AMCC Yes Yes Anritsu Yes Yes Avago Technologies Yes Yes Avvio Networks Yes Yes Bay Microsystems Yes BaySpec Yes Yes BreakingPoint Systems Yes Ciena Yes Yes Cisco Systems Yes CoreOptics Yes Cortina Systems Yes Covega Yes Cube Optics Yes Yes Digital Lightwave Yes Discovery Semiconductors Yes Yes Dune Networks Yes Yes ECI Telecom Yes eGTran Yes Ekinops SAS Yes Enablence Technologies Yes Yes Yes Yes Ericsson Yes Excelight Communications, Inc. --see Sumitomo EXFO Electro-Optical Engineering Inc. Yes Yes EZChip Technologies (Yes) Finisar Yes Fujitsu Yes Yes GigOptix Yes Yes Hitachi Yes Huawei Technologies Yes IDT Integrated Digital Technology Yes Infinera Yes Yes Inphi Yes Ixia Yes Yes JDSU Uniphase Yes Yes Lattice Semiconductor Yes Luxtera Yes Mellanox Yes Micram Yes Mintera Yes Mitsubishi Electric Corporation Yes Monitoring Division, Inc. Yes MorethanIP Yes Yes MRV Communications Yes (Yes) Narda Microwave Yes NEC Yes NetLogic MicroSystems Yes Nistica Yes Yes Nokia Siemens Networks Yes Nortel Yes Yes Oclaro Yes Ofidium (Yes) Oki Optical Components Yes Yes Opnext, Inc. Yes Yes Optametra LLC Yes Yes Optoplex Corporation Yes Yes OpVista Yes Yes Yes Yes Photop Yes Picometrix Yes Santur Yes Yes SHF Communication Technology Yes Yes Yes Sierra Monolithics Yes Yes Sumitomo Electric Industries, Ltd. / Excelight Communications Yes Sunrise Telecom Yes Tektronix Yes Yes Teleoptix Yes Tellabs Yes TeraXion Yes u2t Photonics Yes Yes VI Systems Yes Xelerated Yes Yes Xtera Communications (Yes) (Yes) Yokogawa Yes Yes ZTE Yes (Yes) indicates products in development at time of writing. Source: Light Reading, 2009 Sticky Questions 40 or 100G? 40G is beginning to spread into the wider network, but 100G is still being developed. A question is whether 40G will become no more than a stopgap until 100G becomes commercially available in standardized form. Vendors almost inevitably disagree on the relative status of the two technologies. Tellabs Inc. (Nasdaq: TLAB; Frankfurt: BTLA), for example, has argued that, as 100G has some way to go before it will be properly commercialized (maybe in a couple of years or so), this leaves a considerable window of opportunity for 40G. It is a good solution and will gain considerable momentum, and so will be around for a long time. Further, Tellabs’ systems will allow 40G and 100G to coexist, so that customers will have a migration path. Ciena, on the other hand, has been more skeptical, saying that it is likely that 40G will be deployed only in limited areas of network capacity until 100G is commercially available. NTT America Inc. , for example, has no plans to engage with 40G equipment beyond data center applications of 40G Ethernet, but does expect to deploy 100G Ethernet extensively when 100G Ethernet interfaces become available across the router and switch platforms in 2010. "100G Ethernet, when available, will provide an upgrade path for existing Nx10G links used both in long-haul and intra-POP applications," says Dorian Kim, NTT America's Director of Network Development, Global IP Networks. "For long-haul application, we expect 100G Ethernet to offer a cost-competitive and more substantive upgrade path alternative to 40G OC768 technology. We are working with our transport