OFC/NFOEC 2011: Optical is the only solution for massive bandwidth demand
While reports of the earthquake and tsunami across the Pacific darkened the mood as many of us arrived home following OFC/NFOEC, which was held at the Los Angeles Convention Center on March 6–10, 2011, the news from the conference itself was largely positive. The certainty of continued strong and pervasive bandwidth demand was the underlying drumbeat at the show. Bandwidth growth permeates all aspects of communication networks and simply cannot be supported without reliable, reasonably priced optical components and systems. Noteworthy themes around supporting bandwidth growth included the availability of 100G; plans for 400G and beyond; the need for changes in industry structure, including the advantages of vertical (re)integration; fiber backhaul; and increased coupling and cooperation between the telco and data center worlds. Attendees spanned the communications ecosystem, covering components, chips, FPGAs, systems, data centers, carriers, and investors.
Notably absent were endless panel discussions around the need for FTTx. FTTx is finally here, and while North America is not leading the world, the debate has shifted from the technicalities of PON to simply, “Will it come to my neighborhood?”
100G is here…sort of, mostly
As 100G moves from the lab into initial deployments, a state one attendee described as “technically done,” discussions around 40G are slowly starting to fade. Verizon discussed its 100G link in Europe with 100GE native interfaces using routing equipment from Juniper and DWDM gear from Ciena. The collaboration that moved 100G to deployment much faster than 40G – cooperation within and between standards organizations, including the OIF’s role in crafting an interoperability agreement – meant that all stakeholders had to learn to speak each other’s language. This does not mean, however, that customers want exactly the same solution. There are different needs even within the same segment, such as between Verizon and AT&T, let alone when the requirements of content providers such as Facebook and Google are considered. But when we asked about cooperation, the typical answer was: “Yes, the content providers and telecom carriers are running networks with very different legacy issues, but cooperation leads to vendor interoperability, which leads to lower pricing.”
However, even as 40G discussions seemingly fade, not everything is truly commercially ready in either 40G or 100G. For example, client-side optics remain in flux. The schism on the 100G client side between 10×10 and 4×25 approaches contrasts with the OIF-led industry coordination on the line side. One participant commented that the main difference between 40G and 100G was that at 40G the mess is on the line side while at 100G the mess is on the client side. However, although a 40G SONET/SDH client exists, 40GE and OTN clients are still being debated, and issues relating to expense, density, and even PHY (physical layer) remain, with constituencies sometimes taking positions with religious fervor. As Glenn Wellbrock of Verizon observed in a panel on moving beyond 100G, borrowing liberally from George Santayana, “Those who cannot remember the past (40G) are condemned to repeat it.”
Although the supply chain has clearly matured since last year, there are limited merchant suppliers of some high-speed electronics such as framers and gear boxes. Ovum’s view on the line side was published in the June 2010 report, Optical system and module vendors rush to grab 100G IC expertise. Note that we are also seeing limited optoelectronic suppliers for client-side components such as 10×10, 4×25, and serial 40 transmitters, a subject we will address in forthcoming reports.
Vertical (re)integration – undoing the “unlocking of shareholder value”
Those of us who have remained employed in optical communications over the past decade remember too well the days of “unlocking shareholder value” through component spin-offs and sales of non-essential product lines. Either publicly or not so publicly, we are seeing a movement back to vertical integration. The drivers are clear: time to market, performance, cost, and supply chain control.
We see this vertical integration between different layers of the infrastructure food chain, including Cisco with its 2010 acquisition of CoreOptics and Juniper with its recently announced purchase of Opnext’s coherent intellectual property and software. We also see it within one layer of the food chain. For example, optical networking vendors Ciena, Alcatel-Lucent, and Huawei have developed home-grown 40 and/or 100Gbps transmission systems; Ciena and Alcatel-Lucent are the only two vendors shipping 100G DWDM systems. And recently FPGA suppliers Xilinx and Altera each made acquisitions, of Omiino and Avalon, respectively, in preparation for the next generation of optical transport.
Oclaro’s acquisition of Mintera last year offers a good example of the potential benefits of vertical integration. Both Oclaro and Mintera had developed 40Gbps transmission modules, but with different modulation formats. Mintera demonstrated strong revenue growth of its module, but it had to purchase all the discrete optical components. In contrast, Oclaro had little to no 40G module sales, yet it posted strong sales growth for its internally developed 40G discrete component. The combined company should have a larger total addressable market due to different modulation formats, and should be able to post higher gross margins due to being vertically integrated.   Ovum identified gross margins growth as a major challenge for OC suppliers, and recommended actions to improve performance, in our September 2010 report, In search of sustainable profitability for optical component vendors.
Carriers and content providers are not complaining about the trend towards vertical integration. In fact, it diminishes their concerns, ranging from simple but hard-to-solve constraints such as power consumption and optical budgets, to the survival of a critical single-component vendor in a highly competitive market.
Skittish public company investors and nonexistent VCs
In January 2011, Ovum published ‘Bullwhip Effect’ Destabilizing Telecom Supply Chain, which observed that optical component (OC) vendors had shipped more product than had been consumed by the optical networking vendors. This was consistent with optical components leading the market in coming out of the downturn. We further noted that the 40% growth in OC sales in 2010 would be difficult or impossible to sustain and predicted that 2011 OC revenues would grow more slowly but not shrink. During OFC, Finisar’s stock took a tumble after the company issued a profit and revenue warning due to inventory pileups at some of its customers.
Finisar’s revenue warning was completely predictable based on our research.
At one time, VCs littered the panels and exhibit hall floor, but now they are few and far between. Several optical component and systems start-ups have raised additional capital recently, and a few are looking for their next investors. In fact, analyst Dana Cooperson presented Ovum’s VC data as part of OFC MarketWatch’s “State of the Optical Industry” panel. The data show clearly that VC funding of telecom in general has declined over the past three years. Component and equipment vendors, as opposed to service providers and “others,” claimed 70–75% of all telecoms venture deals in 2008–10. Although it didn’t feel like it at the conferences, optical deals as a fraction of vendor deals actually increased from 5% in 2009 to 8% in 2010. Still, this remained below the 12% we saw in 2008. (Note: see Ovum’s February 2011 report, Telecom Finance in 4Q10: M&A, IPO, VC, and Private Placement Deals, for our latest analysis of venture and other funding deals in telecoms.) “Go strategic” has become the mantra for optical start-ups looking for financing – that is, seek strategic investments from larger, incumbent vendors trying to hedge their bets. While there may be disadvantages to strategic investors, such as right of first offer, existing financial investors are looking for endorsements by strategics, and the vertical integration theme supports this approach.
PON is happening for more than residential broadband: it can be used for backhaul, too
Fact: passive optical networking (PON) is being deployed. PON OLT port shipments exceeded 2 million in 2010 as reported in Ovum’s Market Share: 4Q10 FTTx, DSL, and CMTS Spreadsheet. As noted in the related Market Alert, 2010 FTTx revenues were up 56% to just under $4bn, with almost $2bn coming from PON OLTs, $1.8bn from PON ONT/ONUs, and a little over $200m from P2P. Remember, an OLT port can support 32/64/128 ONUs/ONTs. Instead of debating the merits of EPON versus GPON, PON discussions at OFC were focused on 10G PON deployment plans, what’s next after 10G PON, and the application of PON for wireless backhaul.
Microwave will continue to be used for wireless backhaul, but Ovum’s research shows that over the next several years, fiber’s role as a medium for wireless backhaul will increase substantially. Simply put, fiber can support higher and higher bandwidths and the costs of optical solutions are declining. Optical networking gear and EPON and GPON are being used today for wireless backhaul in the US and China, and 10G PON will be used as it becomes deployed over the next several years. Wireless backhaul was presented as one of three key drivers for 10G PON in Ovum’s February 2011 report, 10G PON Deployment Plans for 2011, and as a key driver for deployment of converged packet-optical products in Ovum’s ON fall 2010 forecast update.
The debate has shifted: OTN vs. MPLS or both?
OTN’s role in the core of the network as a framing protocol is well entrenched, but there was considerable debate at OFC on the role of OTN switching in carrier networks. Carriers with large scale multi-service networks offering private line and wholesale services plan to add a new layer and use OTN switches to manage bandwidth, while others with IP-based next-gen networks prefer label-switched path (LSP) routing using MPLS and MPLS-TP. Juniper, making its first huge splash at OFC, was a key proponent of an MPLS alternative to OTN for core switching.
While the debate rages on, we believe that there is no “one size fits all” architecture and that operators will deploy both MPLS/MPLS-TP and OTN to manage transport bandwidth on their networks. Stay tuned, as this subject is high on Ovum’s upcoming research agenda.
Content providers and carriers: different strokes (and gear) for different folks?
As happened last year, Google and Facebook had a significant presence at OFC. It was quite clear how different the needs of content providers are from those of the traditional telcos. The key differences spring from the relative size and importance of their data centers and their transport infrastructure. Vendors have a challenge trying to satisfy both types of customers, which Ovum will address in much greater detail in subsequent reports.
OFC 2012
By this time next year, traditional carriers, enterprises, and content carriers will have announced many more 100G deployments, and many more systems and component vendors will be showing generally-available gear. 40G will still be shipping, but we expect to see cost compression between the two over the next 12 to 18 months.
Optical component and vendor companies are beginning to develop 400G solutions. We expect to see demonstrations of very early 400G equipment at OFC 2012. One critical element to getting 400G in the field – beyond the 400G systems themselves – is likely to be flexible grid ROADMs and associated components. Multiple vendors are already supporting flexible grid ROADMs,  including Finisar, Nistica, CoAdna, Oclaro, and now JDSU, which announced the addition of liquid crystal on silicon (LCoS) technology to its toolbox to support flexibility. Also note that ADVA announced what’s considered the entire enchilada for the ROADM node, including colorless, directionless, contentionless, and flexible grid. But interestingly, there is no consensus among customers about what flexible grid should look like and how critical it is. At OFC, for example, Verizon was very positive about flexible grid, while AT&T was more cautious given the operational issues flexible grid is likely to create.
Whatever we see at OFC next year, we feel quite confident that bandwidth demands will continue to grow and optics will remain at the center of the solution.
