AT&T: 40G Was Faulty

NEW YORK -- Packet-Optical Transport Evolution -- AT&T Inc. (NYSE: T) has been deploying 40-Gbit/s lines for years, but it hasn't been easy. For one thing, the stuff didn't always work properly.

The point came up early today in Light Reading's Packet-Optical Transport Evolution conference. Speaking during a 100-Gbit/s workshop that preceded the conference sessions, AT&T executive director Margaret Chiosi said she's hoping 100 Gbit/s can avoid the problems that emerged with 40-Gbit/s equipment -- one of those being faulty boards.

"We must have changed out all our cards two times: Class A changes," Chiosi said. "Decades ago, Class A changes would be career-ending. If you had one, you were out of a job. Two, forget it."

AT&T eventually figured out that the problem was industry-pervasive -- other operators kept swapping out equipment, too, and didn't realize the problem was systemic. AT&T traced it down to a single chip, one that was ubiquitous in 40-Gbit/s gear. (Chiosi didn't specify the device any further.)

Obviously, avoiding that kind of thing at 100 Gbit/s would be nice. Vendors are so anxious to differentiate themselves, maybe quality could be a factor, Chiosi quipped.

Another well established problem with 40 Gbit/s was the lack of a standard implementation -- vendors offering different modulation schemes, for instance. In that race to provide differentiation, vendors also kept churning out more cards. AT&T is now on its third or fourth generation of 40-Gbit/s cards, and the continual turnover might be part of the reason vendors have trouble making money, Chiosi contended.

"You put out one version, and it takes us 18 months or two years to certify it and get it out the door -- and once we get it out the door, there's another version," she said.

This problem is already being addressed, at least, with the Optical Internetworking Forum (OIF) developing a 100-Gbit/s framework that can serve as a unified front for the industry.

— Craig Matsumoto, West Coast Editor, Light Reading

Page 1 / 5   >   >>
wentriken 12/5/2012 | 4:35:42 PM
re: AT&T: 40G Was Faulty

But, I thought 40G was the opportunity for vendors to make money by staying off rushing to 100G.

But is seems like all the 40G repeat business is a little too repeat... as in sell the same products many times until it works.

Pete Baldwin 12/5/2012 | 4:35:42 PM
re: AT&T: 40G Was Faulty

40G is still an opportunity, and not just for constant resales --  keep in mind that this story was about what happened earlier in the cycle. The situation's more stable know (at least, I get the impression it is....)

HomerJ 12/5/2012 | 4:35:40 PM
re: AT&T: 40G Was Faulty

Took them a while, but they finally figured out that Frito-Lay was the wrong type of chip.

chechaco 12/5/2012 | 4:35:40 PM
re: AT&T: 40G Was Faulty

Being early adaptor in technological cycle is a gamble. Not only one exposes itself to yet unstable solution but suffers from lack of testing equipment as well. What's more interesting, whether AT&T thinks that their gamble paid off or was waste of resources. "Lessons we've learned from ..." or "Do's and Dont's of adapting cutting edge technology"

Pete Baldwin 12/5/2012 | 4:35:40 PM
re: AT&T: 40G Was Faulty

Ah, someone's brought up the obvious point that i should have noticed: AT&T was very, VERY early in on 40G.  So the 'industrywide' problem, as i described it, didn't paralyze the world or anything - although it did give AT&T conniptions, apparently.

It's still surprising to hear, though.  Pretty big problem to fall through the cracks.

paolo.franzoi 12/5/2012 | 4:35:32 PM
re: AT&T: 40G Was Faulty


I think you have statements that are directly in conflict in your post.  I am going to post some amendments to it that I think clear these up.

You DO have to accept the quality issues that pushing the envelope creates.  What you DON'T have to do is not have an ineffective risk assessment and plan for managing these risks.  Complex products are not possible to be built from the ground up and be completely risk free at introduction. To make this as your stated goal is a complete contradiction to all experience.

What you can do is mitigate the risk and have contingency planning around the most likely risks.  What will fail with the first 100G products will be different than what failed with the first 40G products as EVERYONE will test for that.  But everyone will miss something else and you will be dealing with that fallout.

If you want to mitigate the risk, then you probably need to ensure that you are running parallel processes with multiple technology choices in parallel.  My experience with your company and your bretheren is that your technical staff has been cut so much that this is not a realistic prospect.



mchiosi 12/5/2012 | 4:35:32 PM
re: AT&T: 40G Was Faulty

AT&T is committed to pushing the envelope in technology and solutions if it helps us meet our business needs in a cost effective manner. But we don't have to accept the quality issues which pushing the envelope can create. That is the challenge for this industry which is so HW intensive. To meet the challenges of the market place and still maintain a strong ROI. The learnings in developing 40G products and services must be applied to 100G. So our expectations are very high that most won't be repeated.

hitaoptics 12/5/2012 | 4:35:30 PM
re: AT&T: 40G Was Faulty

        After AT&T changed out all its 40 G line cards two times: Class A changes, why did it not change the linecard vendor? The 40 G line cards vendor obviously had the responsibility for the quality and reliability of its products, and it may loss its credibility by providing the line cards with faulty which influenced many operators.

         When the linecard vendor changed its 40 G linecards for AT&T, the faulty chip may be only one of reasons for the change.  The other reasons might include the concern of its old generation linecard may violate another company’s patent (especially  when new generations are with different modulation formats with old generation).    

Stevery 12/5/2012 | 4:35:29 PM
re: AT&T: 40G Was Faulty

AT&T traced it down to a single chip, one that was ubiquitous in 40-Gbit/s gear.

I assume that's code for sierra monolithics.

paolo.franzoi 12/5/2012 | 4:35:29 PM
re: AT&T: 40G Was Faulty


In these spaces, linecard vendor = systems vendor.  So, what you are in effect saying boot out the system.  You can't go to Alcatel and buy a line card for a Nortel Optical system.  You can't go to Ciena and get a plug for a Fujitsu box.



Page 1 / 5   >   >>
Sign In