100G Ethernet

Infinera Puts 100G Coherent on Pause

Coherent receivers are the optical darlings of the day, but Infinera Corp. (Nasdaq: INFN) is going to take its time getting to them. It will be two to three years before the company produces a coherent photonic integrated circuit (PIC) for 100-Gbit/s transport, a source tells Light Reading. In the meantime, the source says, Infinera plans to offer 100-Gbit/s channels by packing two wavelengths of 50 Gbit/s each -- the trick Nortel Networks Ltd. (now Ciena Corp. (NYSE: CIEN)) used for the first Verizon Communications Inc. (NYSE: VZ) 100-Gbit/s installation. (See Verizon Switches On 100G in Europe.) To clarify: We're talking about transport where each wavelength (or wavelength pair, in Infinera's case) carries 100 Gbit/s. That's different from the aggregate 100 Gbit/s that Infinera's PICs already do, by carrying 10 lanes of 10-Gbit/s traffic. That kind of density has served Infinera well. But operators do want to move to higher speeds, Heavy Reading analyst Sterling Perrin says. So, Infinera's apparent plan goes like this: First, the company will offer its 400-Gbit/s PIC, meaning 10 lanes carrying 40 Gbit/s apiece; this already went through tests last year. (See Infinera Touts 400G PIC.) The source says this will come with the bonus of 25GHz channel spacing, meaning Infinera will carry 160 40-Gbit/s wavelengths per fiber as opposed to the usual 80. After that would come 100-Gbit/s channels using the dual 50-Gbit/s trick. Each pair of 50-Gbit/s wavelengths would fit into one ITU grid slot, meaning Infinera would be able to carry these channels at 50GHz spacing (80 channels per fiber). Finally, the coherent 100-Gbit/s PIC would arrive in two to three years, the source says.

Other vendors are already demonstrating coherent receivers and announcing coherent 40-Gbit/s products. (See CoreOptics Does Coherent 40G, Fujitsu Intros 40G Coherent Optics, and NeoPhotonics Intros 40G/100G PIC.)

Fujitsu Network Communications Inc. , which is using a CoreOptics Inc. 40-Gbit/s coherent module, expects to come out with 100-Gbit/s offerings in 2011, and other vendors at last month's OFC/NFOEC said they expect to match that timeframe. Would Infinera be handicapped by lagging the coherent market? The source who shared Infinera's supposed timetable with Light Reading said no: "They were six years late to 10 Gbit/s, but it didn't matter." The dual 50-Gbit/s option is certainly viable, considering Verizon is using it. Glenn Wellbrock, the carrier's director of backbone network design, defended the technology during his talk at the The Optical Society (OSA) Executive Forum that preceded OFC/NFOEC, saying it works because the two wavelengths fit within one ITU slot. "Spectral efficiency, going into one ITU 50GHz grid, is what mattered to us," he said. Heavy Reading's Perrin thinks Infinera intends to use a multi-wavelength technique for 40 Gbit/s as well.

The problem he sees is that competitors such as Alcatel-Lucent (NYSE: ALU), Huawei Technologies Co. Ltd. , and Opnext Inc. (Nasdaq: OPXT) are already testing out single-wavelength 100-Gbit/s transport, "so the momentum is definitely in that direction," he writes in an email to Light Reading. "It is clear that 100G momentum is around coherent so any vendor will need this ability for long-haul applications. I don’t see a way around that requirement for long-haul DWDM," Perrin continues. Infinera isn't confirming or commenting on its future plans. But Rick Dodd, the company's vice president of marketing, makes it clear that coherent receivers are in Infinera's future. "We absolutely believe in coherent, this is in fact why we opened our design center in Ottawa and hired some of the best guys in the world there," Dodd writes in an email to Light Reading. Infinera considers coherent technology to be necessary only when carriers go beyond having 3.2 Tbit/s of traffic on one fiber, Dodd notes. That would equate to speeds exceeding 40 Gbit/s on an 80-wavelength grid.

Because Infinera's technology is based around the massive integration of components into PICs, Dodd says the company is looking forward to the era of coherent technology, which uses lots of components. — Craig Matsumoto, West Coast Editor, Light Reading

Page 1 / 2   >   >>
ipv5 12/5/2012 | 4:38:41 PM
re: Infinera Puts 100G Coherent on Pause

Could Infinera be required to build 100G (or 400G, 1T) gear as part of the National Broadband Plan?

It would seem that the gear necessary to implement the technology (at a reasonable cost) is required to offer the services.

Whether correct or not, this seems like it should be Infinera's decision.

But how do you get technology into the market at a deployable cost?  Huawei?


Pete Baldwin 12/5/2012 | 4:38:41 PM
re: Infinera Puts 100G Coherent on Pause

Not sure I follow what you're saying.  The National Broadband Plan wouldn't require anything of any equipment vendor, and it's going to focus on the access side of the network, rather than the metro & deep core that Infinera plays in.  I don't think Broadband Plan relates here.

> But how do you get technology into the market at a deployable cost?  Huawei?

Valid question.  The answer I get from most people is that you simply don't -- 100G starts out expensive, with prices gradually lowering as demand and competition increase.

ipv5 12/5/2012 | 4:38:40 PM
re: Infinera Puts 100G Coherent on Pause

I think you are saying that the increases needed are on the edge.  Would there be an increase of load on core infrastructure?

Is that addressed?

Pete Baldwin 12/5/2012 | 4:38:39 PM
re: Infinera Puts 100G Coherent on Pause

There would theoretically be an increased load in the core, but I don't recall that being addressed in the National Broadband Plan.

redface 12/5/2012 | 4:38:38 PM
re: Infinera Puts 100G Coherent on Pause

I wonder what it's like at Infinera these days.  If I am not mistaken, they are still using 10Gb/s On-Off modulation schemes for light modulation while the rest of the world is doing 40Gb/s phase shift keyed modulation which minimizes optical nonlinearities.  While it is nice to say that you have a PIC integrated technology,  it must be pretty hard playing catch-up game all the time.  At this point it should have been pretty clear whether PIC is a blessing or a curse.  The extra difficulties in figuring out how to implement constantly changing technology developments onto an InP chip has resulted in the long lag in Infinera's technology leadership position.  

Pete Baldwin 12/5/2012 | 4:38:38 PM
re: Infinera Puts 100G Coherent on Pause

I'd agree, and I think "long lag" explains the situation well.  The PIC is a long-term advantage but appears to prevent Infinera from getting a jump on any technology shift.

The result is that they're part of the second wave of any given technology. Is that a bad thing? Depending on how big an advantage the PIC is, maybe not.  It worked at 10G, and Infinera might expect it to work at 100G too.

The problem is that it would appear they've got no choice. Whether *they* consider that a problem or not, I don't yet know.

rahat.hussain 12/5/2012 | 4:38:37 PM
re: Infinera Puts 100G Coherent on Pause

interesting that they have just hired some staff to look at coherent technologies; obviously, these technologies will be available in the distant future!

the difference between their 6-year 10gbps delay and this delay is that carriers have standardized on the 100g format, and no amount of hand-waving ("lookee here, we have 75 rails of 1.33gbps", or "hey, we have 12.5 ghz channel spacing, so if we do our fancy math, we have better spectral efficiency") is going to change this basic fact. you can complain about why the standardization was needed so soon, but infn needs to wake up and smell the coffee.

odo <- who remains impressed at infinera's new attempts to market their way out of their technology weaknesses (anyone remember how their oeo devices were called roadms and every reporter bought into that claim?)

paolo.franzoi 12/5/2012 | 4:38:36 PM
re: Infinera Puts 100G Coherent on Pause

"Is it a bad thing?"

Well, maybe...technologies commoditize.  Does all this extra R&D and effort actually result in better financial results?  They have to spend the risk adjusted R&D to hit a later portion of the cycle.  If they are lucky there are still market opportunities to go after.  Also, their competitors have already done their R&D investment and potentially gotten it paid back for already.  If they have, then the competitor can take lower incremental margin and thereby lower prices.

It worked for 1 generation to some extent.  Is it repeatable?  A whole lot less obvious.  It would be interesting to see if they are forced to build some sub-optimal modules to have them in the market before the PIC is available.  If the market moves before the PIC is available, they may get overlaid. 



Balet 12/5/2012 | 4:38:35 PM
re: Infinera Puts 100G Coherent on Pause ...Is not their strongest asset. Lots of ex-Lucent ("I am the only right one") slowly moving, disloyal to their vendors guys, a bit arrogant top brass.

Nice people, just missing common and business sense.
IMHO, Ifinera's glory days are well over.
Pete Baldwin 12/5/2012 | 4:38:32 PM
re: Infinera Puts 100G Coherent on Pause

> Well, maybe...technologies commoditize.  Does all this extra R&D and effort actually result in better financial results?  They have to spend the risk adjusted R&D to hit a later portion of the cycle.

To play devil's advocate - 100G will be an expensive rarity at first.  Maybe Infinera, if they truly are late with coherent, shows up in time for the bulk of the market. Their existing customers will already know about the timing, so those that can afford to wait will wait. Plausible

Page 1 / 2   >   >>
Sign In