Cable Tech

MCI Eyes 40-Gig ULH

DALLAS -- The quest to deploy 40-Gbit/s ultra-long-haul links lives on, and MCI Inc. (Nasdaq: MCIP) says it hopes to have its first commercially deployed 40-Gbit/s network turned on in North America by early next year.

"We're trying to get our 40-Gbit/s network through lab certification this year," says Glenn Wellbrock, director of network technology development at MCI. "I don't know if we will get it deployed, but I hope we do."

Last year, Sprint Corp. (NYSE: FON) announced what it believed to be the first "live-production Internet traffic... carried over a 40 Gigabit link" using gear from Cisco Systems Inc. (Nasdaq: CSCO), StrataLight Communications, and Ciena Corp. (Nasdaq: CIEN). (See Sprint Throttles Up Cisco's CRS-1 and MCI Picks Ciena and Siemens for ULH.) A few days later, MCI and Mintera Corp. announced "the world's fastest Ultra Long-Haul (ULH) connection at 40 Gigabits Per Second -- four times faster than the highest-speed ULH routes deployed today." That milestone was achieved using a long-haul network route spanning more than 1,200 kilometers (see MCI Claims ULH Speed Record).

Wellbrock says MCI has performed three 40-Gbit/s ULH trials in all, and the carrier is inching closer to commercial deployment (see 40-Gig Begins Its Ramp). He says the first trial used equipment from Cisco Systems and StrataLight, while the second trial used gear from Mintera and Ciena. The most recent one used Mintera gear with Optovia Corp. amplifiers -- a setup that uses only about half the amplifiers as found in a traditional long-haul network (see Optovia Doubles Amplifier Reach). Wellbrock says MCI was able to build a link that stretched over 1,600 kilometers with amplifiers placed 160 kilometers apart.

Plenty of other service providers have conducted trials of 40-Gbit/s and even 160-Gbit/s ULH transmission technology. (See, for instance, T-Com Selects Marconi , Lucent Lands Lambdas at Verizon, Carriers Stress Test Their Fiber, and Siemens Claims 160-Gbit/s Milestone.)

But service providers like MCI aren't just concerned with building newer, faster networks. "There are several things we're looking to do with data if we could, beyond just moving it and storing," Wellbrock says. He says that, in the future, MCI could offer services whereby it moves a company's data, stores it, processes it, sorts it, and makes it more accessible based on pre-defined criteria.

"If we just move data, there's a limited amount of revenue we can get out of that," he says.

Wellbrock, whose job entails evaluating new optical and data technologies to support next-generation networks, says MCI has also taken a liking to grid computing (see MCI Goes Ultra Long). He says the carrier has "significantly reduced" the number of servers it is running internally, yet it has increased its available computing power thanks to grid computing, powered by high-bandwidth connections.

The business case for grid computing presents a company like MCI with two challenges, Wellbrock says. First, how will the carrier provide that kind of connectivity? Second, how will the ultra-long-haul network evolve once those connections are made?

Wellbrock says he's scrambling to answer those questions -- and any others that might crop up -- as he prepares to speak at The Future of Optical Networking in Dallas. The one-day conference will evaluate the infrastructure and services that are fueling the evolution of optical networks. The Light Reading-sponsored event will take place on Tuesday, April 12, at the W New York Union Square in New York City, and on Thursday, April 14, at The Fairmont Dallas.

— Phil Harvey, News Editor, Light Reading

Page 1 / 2   >   >>
telsysengineer 12/5/2012 | 3:19:50 AM
re: MCI Eyes 40-Gig ULH Most deployed fiber, new or old, do not have PMD coefficients to be handle 40G over ULH distances. PMD will produce pulse spliting a large fraction of 40G bit and create outages. Does anyone know how the carriers plan to handle PMD over large distance at 40G (or if this not going to be widespread deployment?)
weeshorty 12/5/2012 | 3:19:49 AM
re: MCI Eyes 40-Gig ULH If they use the best fiber available, I calculate that the total DGD including DCMs and amplifiers would be about 5ps. Since a 40G CS-RZ signal has a pmd tolerance of about 8-10ps, this shouldn't be a problem - even less of a problem considering that they wil no doubt be using a strong FEC.
dljvjbsl 12/5/2012 | 3:19:49 AM
re: MCI Eyes 40-Gig ULH The interesting thing I found in this story is that the bulk of the bits that this system will carry will be for machine to machine applications. There was no mention at all of video in any form either for VoD or videoconferencing.
rjs 12/5/2012 | 3:19:47 AM
re: MCI Eyes 40-Gig ULH The total dispersion depends on the distance traveresed (proportional to it) before a 3R operation.


DZED 12/5/2012 | 3:19:46 AM
re: MCI Eyes 40-Gig ULH Excuse me for being really dumb but is this a real 40G?
Is it not 4x10G DWDM, or 16x2.5G DWDM?
This would solve the PMD problems, and be a good deal cheaper, I would think.
lightbridge 12/5/2012 | 3:19:46 AM
re: MCI Eyes 40-Gig ULH I agree with weeshorty that the PMD for the 1600 km link should is about 5ps. In fact, most of the dispersion for this link is in the DCMs - typical is approx 1 ps for a 100km compensator. 16 of these give you about 4 ps PMD (root of the sums squared). The fiber with a link design value of 0.02 ps/sqrt(km) contributes only about 0.8 ps PMD.

However, I don't believe that a 40 Gb/s RZ signal (bit duration: 25 ps) has a PMD tolerance of 8-10 ps. This might be the tolerance for peak DGD (instantaneous value), not for the average DGD. As the DGD randomly fluctuates (of which the time average is called PMD), carriers typically require a DGD tolerance of three times the PMD value to keep the outage time below 20 minutes per year. This would allow for an average PMD value of only 3 ps.
One way to get that low is the use of better DCF - or other ways for dispersion compensation than dispersion compensating fiber.

AutoDog 12/5/2012 | 3:19:44 AM
re: MCI Eyes 40-Gig ULH Dzed,

The StrataLight and Mintera 40G solutions are *real* 40G: single 40G lambdas sent over the long-haul DWDM line system.

(The client side may or may not be a single 40G trib, but that is up to the customer and is independent of the line side.)

As for being more or less expensive than 4x10G, that's the kicker. Time will tell if these 40G vendors can build line cards that can compete with exisitng 4x10 approaches. There are other reasons to consider 40G (fewer lamdas to manage, enables 40G clients (e.g. for HFR), reduced rack space, etc) but cost is the primary driver.

As for 16x2.5G, I'm sure even these new 40G boxes are much more cost competive.

DZED 12/5/2012 | 3:19:43 AM
re: MCI Eyes 40-Gig ULH Thx for the info.

But if PMD is the killer surely 4x10G is a good solution?
yikes_stripes 12/5/2012 | 3:19:41 AM
re: MCI Eyes 40-Gig ULH Each amp stage is going to add around 0.5 ps of PMD. Damn isolators! It's not just the fiber and the DCF you have to worry about.
arch_1 12/5/2012 | 3:19:39 AM
re: MCI Eyes 40-Gig ULH When last I looked, The DWDM Lambda spacing of 40Gbps lambdas was twice the spacing for 10Gbps Lambdas. Is this the case? If so, then goingto 40Gbps doubles the cabacit of a fiber rather than quadrupling it.

When last I looked, distance between amplifiers for 40Gbps was half the spacing needed for 10Gps.

If both of the above are true, then the number of amplifiers for then same BWxMi is the same for 40Gpbs as for 10Gbps. There is no long-haul efficiency gain to justify paying a premium for 40Gbps.

Also when last I looked, the only 40Gbps electronics that you could hook toa 40Gbps laser was an OC768-to-4xOC192 mux. If this is still true, the bitrate for any particular packet is still 10Gbps.

But a year has passed since last I looked. Please advise.
Page 1 / 2   >   >>
Sign In