100G Ethernet

OFC/NFOEC: Why Verizon Wants Terabit

SAN DIEGO -- OFC/NFOEC 2010 -- Equipment vendors might favor 400-Gbit/s Ethernet as the next step after 100 Gbit/s, but Verizon Communications Inc. (NYSE: VZ) would rather jump to Terabit Ethernet, even if it means a five-year wait.

That's because 400-Gbit/s Ethernet doesn't seem worth the potential trouble if 1 Tbit/s can be had. Glenn Wellbrock, Verizon's director of backbone network design, explained his reasoning during a panel at Monday's The Optical Society (OSA) , an adjunct conference to OFC/NFOEC.

"I just don't want to make so many changes in the network, because it's very painful," Wellbrock said. "Once we get a network in place, its really difficult to replace cards that haven't been depreciated, or even to move off that card."

His concerns come from bad experiences in 40-Gbit/s. The industry didn't declare any standards, leading to a variety of approaches that all came with high prices. And the first three years of 40-Gbit/s development were lost on a duo-binary modulation format that Wellbrock considered a dead end.

Verizon talks about 100 Gbit/s so much because "it's the easiest thing to do," he said. The industry has a standard approach that's going to be usable on 10-Gbit/s fiber spans. Verizon has one, possibly two, 100-Gbit/s spans running commercially, based on Nortel Networks Ltd. equipment. (See Verizon Switches On 100G in Europe and Rumor: Verizon's Got Another 100G.)

Standards bodies such as the Institute of Electrical and Electronics Engineers Inc. (IEEE) have yet to settle -- or even formally discuss -- what speed to shoot for after 100-Gbit/s Ethernet. So far, it look as if the carriers want to jump to Terabit Ethernet, while the equipment vendors are favoring 400 Gbit/s -- partly because the chips for a first generation of 400 Gbit/s would be available soon. (See Dare We Aim for Terabit Ethernet?)

Wellbrock said he'd be willing to use 400 Gbit/s if it turns out to be the best the industry can do. But he would like the industry to carefully deliberate whether 1-Tbit/s is doable, rather than go for the "safe bet" at 400-Gbit/s.

"We might be hurting ourselves by making an intermediate step when we don't have to," he said.

Wellbrock can afford to say that, because Verizon has a backup plan: It can always light more fiber for 100-Gbit/s links. Not every span has fiber available, so there might be some leasing involved. But it's something Wellbrock is prepared to do if he runs out of bandwidth before faster speeds are available.

Facebook doesn't have that option. Senior network engineer Donn Lee has been doing the conference circuit for several months, talking about Facebook's imminent need for 1-Tbit/s gear (and emphasizing how screwed Facebook feels about not having 100 Gbit/s available in 2007).

Lee was a speaker at the OSA Executive Forum and repeated his plea -- but with an added hint of resignation. Seeing that neither 400 Gbit/s nor 1 Tbit/s will arrive on his timetable, Lee is ready to build it himself.

"We are gonna have to go to 400-Gbit/s or 1-Tbit/s link aggregation bundles," said Lee. "Call it what you want; it won't be a standard because a standard won't be available. But link aggregation is in the future for us."

— Craig Matsumoto, West Coast Editor, Light Reading

Stevery 12/5/2012 | 4:40:44 PM
re: OFC/NFOEC: Why Verizon Wants Terabit

Verizon is also on

<img alt="" height="335" src="http://www.sticksite.com/spam/spam.jpg" width="335">

Pete Baldwin 12/5/2012 | 4:40:43 PM
re: OFC/NFOEC: Why Verizon Wants Terabit

Spam post has been nuked.&nbsp; Thanks Stevery.

miar70 12/5/2012 | 4:40:37 PM
re: OFC/NFOEC: Why Verizon Wants Terabit

I feel quite sorry for Donn Lee, he has an impossible task it seems. I'm aware he's been out pleading for faster links speeds and aggregated links etc, but he seems to be the only one raising these questions. Therefore the market looks pretty small for any company considering the massive financial investment required to develop it and the fact remains that it's not clear that Facebook is generating enough revenue from all that traffic to pay for such development and deployment...

miar70 12/5/2012 | 4:40:34 PM
re: OFC/NFOEC: Why Verizon Wants Terabit

Well Googles ability to pay is certainly more viable :) There should be a ton of startups all over this, perhaps the VCs just don't see it as interesting enough anymore. Infrastructure is too much of a commodity world and not sexy enough.

Perhaps Donn's repeated message is starting to have an effect on you ;)

Pete Baldwin 12/5/2012 | 4:40:34 PM
re: OFC/NFOEC: Why Verizon Wants Terabit

Google has people hitting the same conferences and delivering the same message, albeit not with the level of specifics Lee presents.&nbsp; And the Internet exchanges, particularly Amsterdam (AMS-IX), are screaming for post-100G links.

I would contend, though, that the market doesn't just look small -- it *is* small.&nbsp; Facebook doesn't need enough terabit links right now to justify anyone building them yet.

Which is actually even more of a reason to feel bad for Donn: He's on an impossible mission, pleading for superfast links that he knows no one's going to bother building yet.&nbsp; He gets two things out of this, though.&nbsp; He can prod the industry to think a little bigger -- to start considering terabit even though 400G seems more feasible.&nbsp; And on the off chance that some weird startup has the Terabit Breakthrough in the labs, they can use Facebook/Google talks as a means of getting funded.

(A confession: I happen to have encountered Donn Lee more often than Google when it comes to this topic, so Lee may be getting a disproportionate amount of ink in Light Reading.)

Sign In