The post-100 Gigabit era
Feature: Beyond 100G - Part 4
The latest coherent ASICs from Ciena and Alcatel-Lucent coupled with announcements from Cisco and Huawei highlight where the industry is heading with regard high-speed optical transport. But the announcements also raise questions too.
Source: Gazettabyte
Observations and queries
- Optical transport has had a clear roadmap: 10 to 40 to 100 Gigabit-per-second (Gbps). 100Gbps optical transport will be the last of the fixed line-side speeds.
- After 100Gbps will come flexible speed-reach deployments. Line-side optics will be able to implement 50Gbps, 100Gbps, 200Gbps or even faster speeds with super-channels, tailored to the particular link.
- Variable speed-reach designs will blur the lines between metro and ultra long-haul. Does a traditional metro platform become a trans-Pacific submarine system simply by adding a new line card with the latest coherent ASIC boasting transmit and receive digital signal processors (DSPs), flexible modulation and soft-decision forward error correction?
Source: Gazettabyte
- The cleverness of optical transport has shifted towards electronics and digital signal processing and away from photonics. Optical system engineers are being taxed as never before as they try to extend the reach of 100, 200 and 400Gbps to match that of 10 and 40Gbps but what is key for platform differentiation is the DSP algorithms and ASIC design.
- Optical is the new radio. This is evident with the adding of a coherent transmit DSP that supports the various modulation schemes and allows spectral shaping, bunching carriers closer to make best use of the fibre's bandwidth.
- The radio analogy is fitting because fibre bandwidth is becoming a scarce resource. Usable fibre capacity has more than doubled with these latest ASIC announcements. Moving to 400Gbps doubles overall capacity to some 18 Terabits. Spectral shaping boosts that even further to over 23 Terabits. Last week 8.8 Terabits (88x100Gbps) was impressive.
- Maximising fibre capacity is why implementing single-carrier 100Gbps signals in 50GHz channels is now important.
- Super-channels, combining multiple carriers, have a lot of operational merits (see the super-channel section in the Cisco story). Infinera announced its 500Gbps super-channel over 250GHz last year. Now Ciena and Alcatel-Lucent highlight how a dual-carrier, dual-polarisation 16-QAM approach in 100GHz implements a 400Gbps signal.
- Despite all the talk of 16-QAM and 400Gbps wavelengths, 100Gbps is still in its infancy and will remain a key technology for years to come. Alcatel-Lucent, one of the early leaders in 100Gbps, has deployed 1,450 100 Gig line units since it launched its system in June 2010.
- Photonic integration for coherent will remain of key importance. Not so much in making yet more complex optical structures than at 100Gbps but shrinking what has already been done.
- Is there a next speed after 100Gbps? Is it 200Gbps until 400Gbps becomes established? Is it 500Gbps as Infinera argues? The answer is that it no longer matters. But then what exactly will operators use to assess the merits of the different vendors' platforms? Reach, power, platform density, spectral efficiency and line speeds are all key performance parameters but assessing each vendor's platform has clearly got harder.
- It is the system vendors not the merchant chip makers that are driving coherent ASIC innovation. The market for 100Gbps coherent merchant chips will remain an important opportunity given the early status of the market but how will coherent merchant chip vendors compete, several of them startups, with the system vendors' deeper pockets and sophisticated ASIC designs?
- Optical transponder vendors at least have more scope for differentiation but it is now also harder. Will one or two of the larger module makers even acquire a coherent ASSP maker?
- Infinera announced its 100G coherent system last year. Clearly it is already working on its next-generation ASIC. And while its DTN-X platform boasts a 500Gbps super-channel photonic chip, its overall system capacity is 8 Terabit (160x50Gbps, each in 25GHz channels). How will Infinera respond, not only with its next ASIC but also its next-generation PIC, to these latest announcements from Ciena and Alcatel-Lucent?
Oclaro: R&D key for growth
“We didn’t sell to Intel,” explains Alain Couder, the boss of Oclaro. “Intel looked for a fab[rication plant] that has good VCSEL technology and that could scale and they found us.”
Couder was talking about how Oclaro became a supplier of vertical-cavity surface-emitting lasers (VCSELs) for Intel’s Light Peak optical cable interface technology. VCSELs are part of Oclaro’s Advanced Photonics Solutions, a division addressing non-telecom markets accounting for between 10 and 15 percent of the company’s revenues.
“I believe very clearly that if a component is available on the market, even if you are a module builder, you are much better off selling to your competition rather than having others do so.”
Alain Couder, Oclaro
Couder joined Bookham in August 2007 and oversaw its merger with Avanex in 2009, resulting in Oclaro. The restructuring has been intensive, with unprofitable product lines discontinued, facilities closed and jobs cut.
“During all this restructuring we never cut R&D,” says Couder. “We have been able to increase our [R&D] people as a third are now in Asia,” he says. “Even in Europe – the UK and Italy – [the cost of] engineers are two-thirds that of the US or Japan.” Indeed Couder says the company is increasing R&D spending from 11 to 13 percent of its revenues. “With growth that we have had - on average 10 percent quarter-on-quarter - we are hiring R&D staff as quickly as we can.”
Vertical integration
Oclaro’s CEO believes being a vertically integrated company – making optical components and modules – is an important differentiator. By designing optical components, Oclaro can drive down cost and tailor designs that it can sell to system vendors and module makers. Such a capability also benefits Oclaro’s own modules.
Couder stresses that there is no conflict of interest selling optical components to module firms that Oclaro competes with. “I believe very clearly that if a component is available on the market, even if you are a module builder, you are much better off selling to your competition rather than having others do so.”
Oclaro supplies components to the likes of Finisar and Opnext, he says, and it has not stopped Oclaro being successful with its 10 Gigabit small form factor (SFF) transponder. Being vertically integrated benefits Oclaro’s modules, growing its market share, says Couder: “Like this year with the SFF and as we expect to be doing next year with our tunable XFPs.” Selling 10 Gigabit-per-second (Gbps) modules also means telecom vendors are buying more modules and less optical components.
“We are going to pursue the same strategy at 40 and 100 Gig,” says Couder. System vendors such as Alcatel-Lucent and Ciena may design their line side optics but as designs become cheaper and performance optimised, Oclaro will be better able to compete. “Our own module solution, or at least our gold box, becomes more competitive than their own design,” he says.
Another important technology aiding vertical integration is photonic integration. “As you put more functions on one chip you get better value,” says Couder. Oclaro has integrated a laser and modulator in indium phosphide that replaces two optical functions that until now have been sold separately. The integrated design takes a third less space yet Oclaro can sell it at a better margin.
40 and 100Gbps markets
Oclaro supplies optical components for 40Gbps differential phase-shift keying (DPSK) modulation and offers its own components and module for 40Gbps differential quadrature phase-shift keying (DQPSK) for the metro/ regional market. Indeed Oclaro is a DQPSK reference design provider for Huawei, the Chinese system vendor with more than 30 percent market share at 40Gbps.
Oclaro is also developing a 100Gbps coherent detection module based on polarisation multiplexing quadrature phase-shift keying (PM-QPSK) modulation, the industry defacto standard. “We think for the very long haul there might be a small market for PM-QPSK at 40Gbps but most of the coherent modulation will be at 100Gbps,” says Couder. “But at the [40Gbps] module level we are continue to be focused on DQPSK.”
Given the recent flurry of 100Gbps coherent announcements, is Oclaro seeing signs of 40Gbps being squeezed and becoming a stop-gap market? “
The only thing I can tell you is that I got this morning again an escalation from one top customer because we can’t supply optical components fast enough for their 40Gbps deployment,” says Couder. “This is all the noise around coherent - 100Gbps will be deployed but even at 100Gbps people are looking at shorter distance solution that are cheaper than coherent. I have not seen any slowing down of 40Gbps.”
He expects 40Gbps to mirror the 10Gbps market which is set for healthy sales over the coming two to three years. Prices continue to come down at 10Gbps and the same is happening at 40Gbps. Ten gigabit modules range from $1,500 to $1,800 depending on their specification while 40Gbps modules are around $6,000. Meanwhile 100Gbps modules will at least be twice the cost of 40Gbps. “There are many sub-networks deployed with [40Gbps] DPSK and DQPSK and I don’t see how operators are going to change everything to 100Gbps on those sub-networks,” says Couder.
Clariphy investment
Oclaro recently announced it had invested US $7.5 million in chip firm Clariphy Communications. Oclaro will develop with Clariphy coherent receiver chip technology for 100Gbps optical transmission and co-market Clariphy's ICs. “We will train our sales force on Clariphy products so we can present to our customers a combination of optical and high-speed components,” says Couder. “We will go as far as giving reference designs.”
In addition to the emerging 100Gbps, there will be marketing of Oclaro’s tunable XFP+ with Clariphy’s ICs and also co-marketing of 40Gbps technology, for example Oclaro’s balanced receiver working with Clariphy’s 40Gbps coherent IC.
Choosing Clariphy was straightforward, says Couder. There were only three “serious” component suppliers: CoreOptics, Opnext and Clariphy. Cisco Systems has announced its plan to acquire CoreOptics while Opnext is a competitor. But Couder stresses that the investment in Clariphy also follows two years of working together.
Couder agrees that the 100Gbps coherent application-specific integrated circuit (ASIC) market is heating up and that there are many potential entrants. That said, he is unaware of many other players that can present a combination of optical components and the ASIC. He also thinks Clariphy has an elegant ASIC that combines the analogue and digital circuitry on one chip.
Meanwhile the Cisco acquisition of CoreOptics is good news for Oclaro. “It took one of the suppliers out of the market; one that was well positioned.” Oclaro is also a supplier of optical components to CoreOptics and to Cisco. “We expect to continue to supply and for us this will be a plus as it [Cisco/ CoreOptics’s solutions] will scale much faster,” he says.
Growth
In other product areas, Oclaro is focussing on its tunable XFP after first launching a extended XFP tunable laser design. “We’re sampling this quarter the regular tunable XFP,” says Couder. “We have been selling a few extended XFPs – the X2 – but the big market is the tunable XFP.”
Oclaro has two offerings – a replacement for the 80km fixed-wavelength XFP that will ship at the end of the end of the year, and a higher specification tunable XFP aimed at replacing 10Gbps 300-pin tunable modules. Couder admits JDS Uniphase dominates tunable XFPs having been first to market. “But we are coming very close behind and what customers are telling me is our performance is better.”
The market for optical amplifiers is also experiencing growth. “We are back to the level before the downturn, back to the level of September 2008,” he says.
The drivers? More optical networking links in the core are being deployed to accommodate growth in wireless traffic, video servers and FTTx, he says. Oclaro is also starting to see demand for lower latency networks. “Some financial applications are looking for lower latency,” he says. “They need gain blocks for 40Gig now and 100Gig tomorrow.” Another telecom segment Oclaro claims it is doing well is tunable optical dispersion compensation modules.
Outside telecom Oclaro's next generation pump products are finding use in cosmetic products while its VCSELs are being used for a future disk drive design. Then there is Light Peak, Intel’s high-speed optical cable technology to link electronic devices. “Intel’s Light Peak will be big; when exactly it will deployed I'm not in a position to say but it will be calendar year 2011.”