40 Gbit/s Takes the Slow Lane

BALTIMORE — NFOEC — Executives from several top optical networking companies splashed cold water on 40-Gbit/s technology at an investment conference held here yesterday at the NFOEC (National Fiber Optic Engineers Conference).

A panel of executives told a packed house of investors, engineers, and fellow vendors that they have pushed back their estimates for widespread deployment of the 40-Gbit/s technology until at least 2003 or 2004.

“We’ve been aggressively developing 40 Gbit/s,” said Ira Deyhimy, vice president of product development for Vitesse Semiconductor Corp. (Nasdaq: VTSS), at the conference organized by CIBC World Markets. “We thought 40-Gbit/s sales would ramp up in late 2001. Now we’re saying it will be more like late 2003 or 2004.”

The 40-Gbit/s question seemed to dominate discussion during two panel discussions centered around optical components and optical networking systems.

Top executives from some of the leading component companies in the industry — including Fred Leonberger, CTO of JDS Uniphase Inc. (Nasdaq: JDSU; Toronto: JDU); Frank Levinson, CTO of Finisar Corp. (Nasdaq: FNSR); Ira Deyhimy of Vitesse; and Henry Yaffe, CTO of Yafo Networks — voiced the general consensus that the uptake for 40-Gbit/s technology has slowed considerably from what they had once predicted.

The sentiment was echoed by optical system providers during another panel discussion. Elizabeth Perry, senior vice president at Ciena Corp. (Nasdaq: CIEN), said she doesn’t expect high volume shipments of 40-Gbit/s technology in Ciena long-haul products for some time. And Chris Hamilton, chief technologist, from Williams Communications Group (NYSE: WCG), the only carrier involved in any of the panel discussions, said he doesn’t yet see a need for 40-Gbit/s link deployments in most stretches of his company’s network.

There are several factors behind the delay in deployment. For one, developing new high-speed technology is never easy. At 40 Gbit/s there are many complications associated with the packaging of components and the physics involved in such high-speed transmission.

“At 40 gig there is an inescapable dispersion problem,” said Yafo's Yaffe. “We can probably deal with it, but at what cost? The cost equation really becomes critical.”

Dispersion occurs when a light signal spreads out over distance. These signals typically must be adjusted using special components. The problem grows particularly acute at higher speeds such as 40 Gbit/s.

While vendors like to tout faster and faster speeds, the reality is that many service providers haven’t completely adopted 10 Gbit/s and are still buying huge amounts of 2.5-Gbit/s gear. For example, Nortel Networks Corp. (NYSE/Toronto: NT) was one of the first companies to come out with 10-Gbit/s solutions, but now the company’s sales of 10-Gbit/s gear are beginning to flatten, say some analysts. Meanwhile, Ciena has actually seen a ramp-up in its OC48 (2.5 Gbit/s) sales. One reason is that carriers like Sprint Corp. (NYSE: FON), for example, haven’t moved to 10 Gbit/s as quickly as some had expected. And with 2.5-Gbit/s prices continuing to fall, it makes lower-speed technology an even more attractive choice in this capital constrained market.

But that doesn’t mean that component makers are completely sidelining their 40-Gbit/s developments. Fred Leonberger, of JDSU, says that his company can’t afford to put 40 Gbit/s on the back burner, because OEM customers are already looking for samples.

“Those of us working on 40-Gbit/s technology are still feeling pressure from OEMs,” he said. “They still have a clear vision of when they want the technology for their prototypes.”

While long-haul deployments of 40 Gbit/s may be two to three years away, short-reach applications are likely to be in customer trials sometime in the second half of next year, said Elizabeth Perry of Ciena. Analysts agree.

“Absolutely, 40-Gbit/s technology will be a reality quicker in short-reach applications like terabit routing than in long-haul applications,” says Jim Jungjohann, of CIBC. The reason? As IP data grows in the network core, carriers can use 40-Gbit/s links to hook multiple core routers together.

“The interconnect piece is where we will likely see 40 Gbit/s first,” said Perry in an interview after the panel discussion. “It makes sense, because you can use a VCSEL [vertical cavity surface emitting laser] array to make a cost-effective solution — and because the distances are so close, that isn’t an issue.”

The revised outlook for widespread 40-Gbit/s technology could affect many a startup that raised money on the premise that the market would develop sooner. Venture capitalists say that the life expectancy for most startups has been greatly reduced because of the lack of funding.

“Over the next 15 to 18 months, I see about 75 to 80 percent of these companies going away,” says Melissa Crane, a partner at VantagePoint Venture Partners. “The capital market is just so tight these days. The only hope some of these companies have is for a bigger company to buy them — because there just isn’t an IPO market right now.”

- Marguerite Reardon, Senior Editor, Light Reading

<<   <   Page 2 / 3   >   >>
doublebait 12/4/2012 | 8:06:15 PM
re: 40 Gbit/s Takes the Slow Lane Small thinking only leads to small ideas.
Half-Inch Stud 12/4/2012 | 8:06:14 PM
re: 40 Gbit/s Takes the Slow Lane Long-Haul is unfavorable to 40Gig signaling for more than one compelling reason; Comparative Hardware costs, Signal/Noise compromise at Modulation, Clock Jitter, and the dispersion.

Sure, Dispersion is the show-stopper in and all-bad arguement. Besides, Edge card strategy now tracks the ISP strategy; build as you grow, rather than overbuild and wait. Upgrades should only occur when they have to.

Upon entering the COs, the Edge/Core, Router/Switch has to DeMux the 40Gig string anyway...Remind me again, who wants to do all that? And for what functional benefit?

"Wax On....Wax Off"
"But Why??"
"You no ask. you wax-on wax-off...like this"
"I'm done"
"OK, tomorrow you paint all this."
"But Why?!"
"you no ask....

40gig, it builds character, nothing else."

H.I. Stud

realguy 12/4/2012 | 8:06:13 PM
re: 40 Gbit/s Takes the Slow Lane $1M for PMD compensator! Frankly I think this is cheap! if it really solves the problem. PMD is not quite understood anyway. If PMD mixes with nonlinearities in the fiber, there is no way the signal can be corrected. The only solution to PMD is becoming smart and avoid it all together. Stay at 10G and 2.5G and relax.
gpearson 12/4/2012 | 8:06:13 PM
re: 40 Gbit/s Takes the Slow Lane The dispersion problems are more or less an LD issue. In Metro networks, 40-Gig will save a good bit of interface real estate. Floor space has a cost, too.

Assumes, of course, there is an ecomony of scale to make 40-Gig pipes a requirement in the Metro space. For the short term, I think LR called it right. 40-Gig will not be ready for prime time in the near term.

By the way, in a conversation with a supplier working on 40-Gig, the suggested cost for an adaptive PMD (polarization mode dispersion) compensator was close to $1M. Of course, that's a prototype cost, not production. But, it does emphasize how far 40-Gig has to go before it becomes commercially viable (at least for the LD market.)
realguy 12/4/2012 | 8:06:12 PM
re: 40 Gbit/s Takes the Slow Lane Why? because I said so!
mma 12/4/2012 | 8:06:05 PM
re: 40 Gbit/s Takes the Slow Lane 10-Meg Ethernet, 100-Meg Ethernet, 1-Gig Ethernet, 10-Gig Ethernet ...

Doesn't 100-Gig come next?
plcornet 12/4/2012 | 8:06:04 PM
re: 40 Gbit/s Takes the Slow Lane Ok you brainiacs. It boils down to cost. If it is cheaper to pack it tighter, thts the way the industry will go. If 40g is cheaper, then thats the direction.

For long haul applications, the other factor is which one will be easiest to integrate into existing networks. Obviously, there is not going to be very many coorporations building any new networks for awhile (reference 360), so the big push in today's market will focus on existing networks.

Inregards to Metro, reference the 1st paragraph.

Both technologies are valid, and both will eventually be used in accordance with the different characteristics of each technology. One will not win out against the other. They will just be used in different applications.
Superman 12/4/2012 | 8:06:03 PM
re: 40 Gbit/s Takes the Slow Lane Does anyone know how clock and data recovery should work on 40G? Is it going to be PLL or DRO method? Who is currently working on this single module?
fhe 12/4/2012 | 8:06:03 PM
re: 40 Gbit/s Takes the Slow Lane 40G is dead? Come again?

You guys were probably the ones who said 28.8 was an overkill when you were using 9600; or you say broadband access is more than enough when you are *STILL* using 14400 modems *currently*...

voyeur 12/4/2012 | 8:06:02 PM
re: 40 Gbit/s Takes the Slow Lane Phos,

I am not one to quibble about the waste of time and money on 40G now. I do believe it will be needed but I don't think the takeup will accelerate until a few years from now.

I do have an issue about what you say about DSL. The main proponents were the CLEC's, since they could get more services through that route. But since they didn't have local access to the customer (read - "Had to go through the RBOC's"), they would never get it to pay off. And for the RBOC's, they basically had no incentive to roll out DSL. Their mantra "Why offer DSL when we already offer T-1, etc at higher prices?" Until there is true competition at the local level, where RBOC's compete against RBOC's, 40G, 10G Ethernet, FTTH, xDSL, etc will be pipedreams for the common folk. I am kind of hoping the cable folks can light a fire under them (but then again, they are monopolists, too)
<<   <   Page 2 / 3   >   >>
Sign In