& cplSiteName &

Dutch Exchange Rolling Its Own 100G

Craig Matsumoto
2/17/2012

SANTA CLARA, Calif. -- The Amsterdam Internet Exchange B.V. (AMS-IX) is designing its own 100Gbit/s module for metro reaches, because it can't wait even nine months for commercial modules to hit the market.

Modules for the 100GBase-ER standard, which covers spans of 40km, should be available by the end of the year, but AMS-IX needs something next quarter, says Martin Pels, an AMS-IX senior network engineer. Pels spoke Thursday on a panel at a Technology Exploration Forum, an ongoing series of events put on by The Ethernet Alliance for discussion of next-generation requirements.

As an Internet exchange, AMS-IX needs to move enormous amounts of traffic, making it an early adopter for technologies such as 100Gbit/s transport. (See AMS-IX Sticks With Brocade.)

And it's using a lot of 100Gbit/s links, relatively speaking. AMS-IX's plans for 2012 involve adding two switches to its network core (bringing it to a total of six) and, judging from a diagram Pels showed, at least a couple dozed 100Gbit/s links in various places around the network.

The AMS-IX kludge starts with a 100GBase-LR4 module -- a type that's commercially available but is only meant for 10km spans. In front of that will be added a discrete semiconductor optical amplifier (SOA) or praseodymium-doped fiber amplifier (PDFA), which is awesome because we get to mention praseodymium for the first time ever on Light Reading.

For the latter case, AMS-IX would just buy a PDFA commercially; they're available in 1-rack-unit boxes, Pels tells Light Reading. The SOA is a custom chip that a vendor is developing for AMS-IX.

It's possible the AMS-IX module won't be used for very long -- but then again, it's possible the first ER4 modules will be expensive enough that AMS-IX will stick with its own module for a while longer, Pels says.

And who knows -- maybe other customers will want to try AMS-IX's module. That's kind of how the 10x10 multisource agreement for 100Gbit/s modules got started. (Not really, but kind of...)

— Craig Matsumoto, Managing Editor, Light Reading

(8)  | 
Comment  | 
Print  | 
Newest First  |  Oldest First  |  Threaded View        ADD A COMMENT
SimonParry
SimonParry
12/5/2012 | 5:42:21 PM
re: Dutch Exchange Rolling Its Own 100G


"which is awesome because we get to mention praseodymium for the first time ever on Light Reading."


As opposed to when you mentioned it in 2002: http://www.lightreading.com/do...

DCITDave
DCITDave
12/5/2012 | 5:42:19 PM
re: Dutch Exchange Rolling Its Own 100G


Thanks. Good to see our search engine is working.

Pete Baldwin
Pete Baldwin
12/5/2012 | 5:42:19 PM
re: Dutch Exchange Rolling Its Own 100G


Nope, I saw that one, and it doesn't count!  It's got the acronym but doesn't name the element!

Duh!
Duh!
12/5/2012 | 5:42:17 PM
re: Dutch Exchange Rolling Its Own 100G


We had 1310nm and 1490nm SOAs from at least three different vendors in our lab 3 or 4 years ago.     Did they say why they needed to have these devices custom fabricated?

BigBro
BigBro
12/5/2012 | 5:42:17 PM
re: Dutch Exchange Rolling Its Own 100G


All these 100GBASE-xR4 standards run 4 links at 25Gbps in parallel, so each link can have different jitter, etc.


Is extending the range really as simple as just juicing up the signal? Does amplifying it not introduce additional jitter, which, if different between the lanes, could be a big deal?

Pete Baldwin
Pete Baldwin
12/5/2012 | 5:42:15 PM
re: Dutch Exchange Rolling Its Own 100G


> Is extending the range really as simple as just juicing up the signal? Does amplifying it not introduce additional jitter, which, if different between the lanes, could be a big deal?


I would think it does. Then again, AMS-IX might not need all 40km, so maybe the solution doesn't have to be perfect...?


The setup would also be clunkier than a module by definition (i.e., bigger, you have to go through the trouble of makingit, etc...). Pels made it sound like they'd be happy to do things this way as long as it's cheaper than an ER4, but that might just be posturing on his part.

Pete Baldwin
Pete Baldwin
12/5/2012 | 5:42:15 PM
re: Dutch Exchange Rolling Its Own 100G


You know, that kind of puzzled me too, after the fact. I should go back and ask.

^Eagle^
^Eagle^
12/5/2012 | 5:42:14 PM
re: Dutch Exchange Rolling Its Own 100G


actually if you use a properly designed SOA, you do not introduce any additional jitter.  For the SOA it is a question of where on it's operating curve you run it.  How close to PSAT is it?


if the SOA has enough gain, and correct threshold, and enough Pout, then it should be a nice solution.  You want to stay in the iinear regime for the SOA and not run it at saturation (counter to what you normally do with an EDFA or PDFA)


But I agree with the post.  There are several companies making off the shelf SOA's that are fully packaged.


We did an experiement with 16 and 32 streams of 10gig going through an SOA back in 2002 / 2003.


2x25 should relatively straightforward.  


sailboat

Featured Video
Upcoming Live Events
October 22, 2019, Los Angeles, CA
November 5, 2019, London, England
November 7, 2019, London, UK
November 14, 2019, Maritim Hotel, Berlin
December 3-5, 2019, Vienna, Austria
December 3, 2019, New York, New York
March 16-18, 2020, Embassy Suites, Denver, Colorado
May 18-20, 2020, Irving Convention Center, Dallas, TX
All Upcoming Live Events