The Lightwave Blog

Posted by Stephen Hardy

Today's news of Juniper's OEM agreement with IBM puts me in mind of the switch/router vendor's link with Nokia Siemens Networks (NSN) for an integrated approach to IP over DWDM -- a concept practically owned by Cisco.

I spoke earlier this month to Pathmal Gunawardana, director of business development at NSN, who explained why the combination of NSN and Juniper could beat Cisco at its own telco network game. While he declined to make a feature-to-feature comparison of Juniper's boxes with Cisco's, he was much more willing to suggest that NSN provided the stronger DWDM portfolio.

Specifically, said Gunawardana:

• NSN has a broader, more scalable DWDM offering. While Cisco's gear is principally targeted at the metro, in Gunawardana's view, NSN can offer metro, long-haul, and ultra-long haul platforms, he said.
  
• Since NSN was the market leader in 40-Gbps sales in 2008, the company would be better able than Cisco to leverage volume to create a cost-effective 40-Gbps interface -- not to mention a future 100-Gbps interface.
  

Speaking of these interfaces, Gunawardana said that the 40-Gbps interface would likely leverage the company's current DPSK offerings, at least initially. The company has targeted what he called "CP-QPSK" -- for "coherent polarized QPSK" -- as the technology of choice for the 100-Gbps interface. CP-QPSK will be similar, if not identical, to the dual-polarization QPSK with coherent detection the OIF is currently working on. Once perfected, the technology would likely be used for 40 Gbps as well, he predicted.

Gunawardana said that both interfaces should be available within the next 18-24 months, although he hinted that the 40-Gbps capability would likely appear in the early part of that window and the 100-Gbps technology toward the end. Since the interface will be integrated into the Juniper hardware, Gunawardana conceded that it could be used with DWDM equipment not from NSN. However, such a pairing would likely fall short of the performance a Juniper/NSN combination would provide, he asserted.

Labels: 100G, 40G, Cisco, Juniper, Nokia Siemens Networks, NSN

Posted by Stephen Hardy

Speaking to a group of media and analysts after last Friday's OIF interoperability demo at Verizon's Waltham, MA, facility, the carrier's vice president of network architecture, Stuart Elby, said he expects that at least one of his current vendors, and perhaps as many as three, will deliver tenable 100-Gbps networking platforms by the end of this year. He added that he expects the platforms will be based on technology "like" the dual-polarized QPSK with coherent detection around which the OIF has rallied the industry (including Verizon), saying he believes the platforms will be "as close to that as exists" at the time.

Elby also said he expects that Verizon will deploy some of the equipment it receives, but not in large numbers. As was the case with 40G, he expects the first generation of 100G platforms will be extremely expensive, and greater deployment will wait until further iterations of the technology reduce 100G's price tag. He said he had doubts that 40G prices would ever reach a level 2.5 times that of 10G, partly because the price of 10G technology continues to shrink. He said that 100G might enjoy a more aggressive downward cost run than 40G, due to greater deployment in data center environments.

Labels: 100G, 40G, Verizon

Posted by Stephen Hardy

The folks at StrataLight Communications followed up the Cisco/Comcast 100G demo announcement by confirming that they had provided the optics. But according to Ross Saunders, general manager of next-generation transport at StrataLight, the company isn't focusing on productizing the interface used in the demo in the short term.

Saunders wouldn't provide many details (such as modulation format) of the "DWDM optics" his company supplied for the demo, other than to confirm that it involved a single wavelength running at 100G. The interface did not use the 100G chipset the company recently announced, either.

As the IC announcement implied and Saunders confirmed, StrataLight is focusing its 100G attention in the near term on a muxponder product that could be used to aggregate 10G and 40G wavelengths as an interim step toward serial 100G transmission. Saunders says the muxponder or "multiplexing transponder" is slated to be delivered to an OEM and a carrier for trials within the next month. When and if the prototype evolves into a commercially available product will hinge largely on the reaction it generates during these trials.

Meanwhile, the company continues its work in the 40G space.

Labels: 100G, 40G, Cisco, Comcast, StrataLight

Posted by Stephen Hardy

Quickly reviewing my reporter's notebook while I chomp on a bagel in the press room Thursday morning, here's what I find:

Net Insight, which has ridden its DTM technology to success in the video broadcast market, hopes to make a bigger name for itself in North America through support of telepresence and multiservice delivery...Mintera execs Terry Unter (pres and CEO) and Niall Robinson (VP, prod marketing) claim that they've shipped their 40G transponders to "seven or eight" customers, five of whom have designed the module into a system -- and, in some cases, more than one...Speaking of 40G, rumor has it that the "high speed" company Opnext currently is looking at most closely for acquisition (see blog post below) is StrataLight Communications...Having speculated in the same blog post that it might make sense for Opnext to look at CoreOptics, I asked Saeid Aramideh, VP of global sales, marketing and business Development, about the possibility. He said his company sees itself as a consolidator, not a consolidatee -- and hinted that CoreOptics might prove that in the reasonably near future...Everyone is talking about having an RF return path capability on their PON or other FTTX product to address the MSO market as well as other carriers who want to add on-demand services without abandoning their current RF infrastructure. EPON vendor Alloptic claims that at least some of the GPON vendors touting this capability are using their technology -- which they refuse to sell to other EPON vendors...For every vendor touting WDM-PON at the show (and there are several), there's another saying that there probably isn't much of a market for it between now and the time more affordable 10G PON technology will be available...While the OIF has given its blessing to dual-polarization QPSK for 100G transport, neither Hitachi Telecom nor Adva Optical Networking think that's the right way to go. They don't seem too concerned about the OIF's current direction. If AT&T; or Verizon announced they'd prefer a different modulation format, the OIF would undoubtedly pull an about-face, one source speculated.

More to come later...

Labels: 100G, 40G, NXTcomm, PON, WDM-PON

Posted by Stephen Hardy

As yesterday's funding announcement pointed out, CoreOptics plans to use part of its $25 million capital influx on new product development. In the short term, that means 40G technology; according to CoreOptics VP of Global Sales, Marketing and Business Development Saeid Aramedeh, a new IC is expected to be announced in the next few weeks, to be followed by the second half of next year by a 40G transponder. A 100G product will follow. The trick for CoreOptics, he says, is timing the 100G introduction in time to meet market requirements without killing the 40G offering's prospects.

The company plans to pair its MLSE-2 electronic dispersion compensation technology -- yes, it's based on maximum likelihood sequence estimation and it's the company's second version of the technology -- with coherent optical receiver/detection formats. (Coherent receiver technology also is being used by Discovery Semiconductor and Nortel, the latter pairing it with dual-polarization quadrature phase-shift keying.)

CoreOptics already offers 40G technology -- it was designed into the platforms Marconi (now part of Ericsson) sold to Deutsche Telekom -- but this part of the company's business represents only about 10% of sales. Aramedeh naturally expects this to grow, albeit slowly.

"Personally, I don't see this happening over the next six months or 12 months. It's probably more of a two-year journey before we see 40G in mass deployment from our perspective," he says.

Which brings us to the dilemma 100G poses. "Certainly if you look at the architecture of 100-gig, it has much promise for being more cost-effective than any 40-gig solution," Aramedeh explains. "So the question becomes if we were to expedite 100-gig delivery, would that make sense and then 40-gig would have a shorter life or never go into mass deployment."

A short tenure for 40G would pose a problem for CoreOptics, Aramedeh implied. "When we look at technologies that are needed to make 100-gig a reality, electronic dispersion compensation is definitely a part of that. And for us, the technology roadmap to support 100 gig goes through 40 -- we have to do 40 and 40 coherent to get the learnings from the experience before we can confidently say what architecture we could implement for 100-gig transmission," he explains.

For now, Aramedeh thinks he'll have enough time with 40G to get 100G right. "From a timing perspective, certainly we don't see that [100G] as an immediate need. And the market is divided on that," he admits. "You talk to some folks and they say, 'You need to go into trials the end of next year.' But for volume ramp, we certainly don't see it before 2011 or 2012."

Labels: 100G, 40G, CoreOptics

Posted by Stephen Hardy

In light of today's announcement regarding the partnership between JDSU and Mintera, it's clear JDSU believes the 40G market is worth pursuing. This belief comes from the company's insight into the RFPs their system house customers are fielding, according to Director of Product Marketing Craig Iwata.

"JDSU is involved in pretty much every transport RFP that's out there. And they're all coming through with requirements for 40-gig," he told me.

While there's a difference between requiring 40G capability and actually fielding it, Iwata also told me that the market research community is backing this viewpoint as well. As part of the presentation he and Mintera's Niall Robinson delivered to me last week, Iwata said that when you look at the 40G line-side interface forecasts from CIR, Infonetics, and Ovum-RHK, the average unit CAGR for 2008 through 2011 is 100%, with their respective hockey sticks all taking a distinctly upward turn beginning in 2009.

Neither Ovum-RHK nor Infonetics have much on the public parts of their websites covering their 40G forecasts. You can find a little bit about CIR's view here and here.

As usual, the big issue will be cost, Daryl Innis of Ovum-RHK told me. "There's still a serious price problem. The cost of a 40-gig line card is something like five to six times the cost of a 10-gig line card. Until that cost comes down, that's going to limit the growth rate of 40-gig," he said.

I think there's mounting evidence that there will indeed be a market for 40G technology, particularly in the long haul and ultra long haul. (Which probably translates into a "good" but not "great" market.) Certainly 100G will eventually overtake 40G in terms of popularity, but that's going to be a fairly long time coming. The standards have to be put in place first -- and if 40G technology currently is 5X to 6X your average 10G system, one can only imaging the price differentials of the first and even second generations of 100G platforms.

Labels: 40 Gbps, 40G, JDSU, Mintera