x
100G Ethernet

100G Watch: OFC/NFOEC Warmup

At OFC/NFOEC next week, 100-Gbit/s optics will become an even hotter topic than they've been in the past several months. Here's a look at where some of the discussion has been headed so far, from the service provider perspective.

  • When Verizon Communications Inc. (NYSE: VZ) starts deploying new fiber routes, they'll be built with 100-Gbit/s transmission in mind, Glenn Wellbrock tells analyst Roy Rubenstein on the Gazettabyte blog.

    Wellbrock, director of backbone network design, expects those build-outs to start by 2012, possibly earlier. But here's the catch: "100 Gbit/s" here implies coherent detection, which in turn means Verizon Business won't use any optically based dispersion compensation. Dispersion will be corrected using the receiver's electronics.

    That's likely to be the norm for long-haul 100-Gbit/s links, anyway. But it means that Verizon will insist on coherent detection for 40-Gbit/s traffic, too, on those new fiber routes.

  • No one expects the big customers to stop rallying for speeds beyond 100 Gbit/s. Donn Lee of Facebook has been doing so for months, going to various industry conferences to say his company needs 100-Gbit/s Ethernet already. Light Reading reported on his appearance at a seminar organized by The Ethernet Alliance , where his talk touched off a discussion about Terabit Ethernet. (See Facebook: Yes, We Need 100-GigE.)

    He reprised that talk in early February and brought up the topic again a few weeks ago, at the Ethernet Technology Summit in San Jose, Calif.

  • Along possibly similar lines, OFC/NFOEC's Service Provider Summit will be keynoted by Vijay Gill, senior manager of engineering and architecture at Google (Nasdaq: GOOG). The subject is "Building Very Large-Scale Computer Infrastructure," and it's ostensibly about the computers themselves -- although it's a good bet 100-Gbit/s Ethernet, or some bigger number, will get a mention.

    Other recent bits and pieces from the 100-Gbit/s world:



    — Craig Matsumoto, West Coast Editor, Light Reading

  • Page 1 / 3   >   >>
    wentriken 12/5/2012 | 4:40:49 PM
    re: 100G Watch: OFC/NFOEC Warmup

    More data: http://it.slashdot.org/story/1... (terrible source, I know)


    Google is made up of 500,000 systems, 1 million CPUs and 1,500 gigabits per second (Gbps) of bandwidth, according to cloud service provider Neustar. Amazon comes in second with 160,000 systems, 320,000 CPUs and 400 Gbps of bandwidth, while Rackspace offers 65,000 systems, 130,000 CPUs and 300 Gbps.


    I found three more non-ISP customers!

    nelson3748 12/5/2012 | 4:40:52 PM
    re: 100G Watch: OFC/NFOEC Warmup i should have been more specific, i meant VoD and game distribution, not gaming
    Pete Baldwin 12/5/2012 | 4:40:53 PM
    re: 100G Watch: OFC/NFOEC Warmup

    > Regardless, OFC is a technology conference, so my expectation is that the focus should be on these sorts of forward looking projects.  Personally, I much prefer this topic to the all optical network and optical switching which I think is a real boondoggle, but that is a post for another day.


    george - i'm with you there.  The marketing around 100G might get annoying, but there's certainly legitimate reason for the technical sessions to talk about it (or about 400G or Terabit).


    What do people think of the technical sessions at OFC/NFOEC?  Still good?


    Pete Baldwin 12/5/2012 | 4:40:53 PM
    re: 100G Watch: OFC/NFOEC Warmup

    > In short, I hope the struggling telecom vendors have not invested the sink on a cool technology that will have minimal volumes, at the very best of time but we all know that many already have done this exactly.


    There's probably some truth to that.  And volumes certainly won't be big at first.


    But with carriers claiming to need 100G so badly, it's a way for a company, big or small, to try to get its foot in the door for the next several years.  Many of them had to try. 


    Pete Baldwin 12/5/2012 | 4:40:53 PM
    re: 100G Watch: OFC/NFOEC Warmup

    You know, I thought so too, a couple of years ago.  But the video game guys work hard to make sure you don't need that much bandwidth to play.  What I've been told is that your PC is still the gating factor, which is why gamer PCs are a viable market but gamer ISPs aren't.


    I still think there's potential for that to change, but it still might be a ways off.


    Distribution could make use of wide pipes.  I think it'll be dwarfed by video (read: TV, VoD) though.

    nelson3748 12/5/2012 | 4:40:54 PM
    re: 100G Watch: OFC/NFOEC Warmup won't the video & game distributors need some really wide pipes?
    furious_george 12/5/2012 | 4:40:56 PM
    re: 100G Watch: OFC/NFOEC Warmup

    @wentriken 


    I don't believe I am in disagreement with your point.  The overwhelming majority of the world is perfectly content with 1GE.  Further I agree that "This it-would-be-nice-if-it-was-available-and-cheap demand does not appear to me to support a healthy ecosystem."


     


    However, so what?  See page 7 of http://www.ieee802.org/3/hssg/public/apr07/hays_01_0407.pdf.  This presentation was made back in April'07 and is predicting 40GE at the server in 2011 or 2012.  In the meantime lots of companies have been pushing 40G interfaces and some have announced revenues.  I think it is perfectly natural that some companies will want to push 100G to overcome weakness at 40G, or to excite investors, or to continue beating a technology drum.  Of course 100G will not be a huge nor profitable market except for <three players for the next >three years.  However, this is the prerogative of those companies and their respective boards.


     


    As you point out with the short reach application, 100G will be a niche in a few special use cases.  I think there are other use cases beyond short reach, for example the announcements by Verizon and AT&T point to interest beyond the data center and are targeted at >10km reach as are most of the needs of the large internet exchanges. Furthermore, I expect to be inundated with marketing details on 100G potential applications at OFC.  We will have to wait and see what comes to fruition.


     


    Regardless, OFC is a technology conference, so my expectation is that the focus should be on these sorts of forward looking projects.  Personally, I much prefer this topic to the all optical network and optical switching which I think is a real boondoggle, but that is a post for another day.

    laserboy 12/5/2012 | 4:40:56 PM
    re: 100G Watch: OFC/NFOEC Warmup

    100G is just another expensive hero experiment that the R&D folks like to tout with no solid business case behind it. The 10G market exploded in the perfect storm: lots of available venture capital, the birth of the internet that was for the most part at the time only used by schools and geeks, emerging eCommerce and a very robust economy coupled tightly with lackadaisical financial reporting rules in place at the time allowing for inflated revenues and profit margins.


    Today, everyone who wants an internet connection, has one. Everyone who wants a cell phone, has one. The download speeds of today suffices for web browsing and email that are the major reason for an internet connection for the vast majority of the population.


    So where is the market for 100G? It will only take hold when it is far cheaper than transporting today's/tomorrow's VoIP/data traffic over today's 10x10G or 3x40G solutions. This will not be the case for years to come so the lower volume on 40G today ($100M market worldwide as of recent HeavyReading market reports) will be even lower tomorrow on 100G.


    In short, I hope the struggling telecom vendors have not invested the sink on a cool technology that will have minimal volumes, at the very best of time but we all know that many already have done this exactly.


     

    ^Eagle^ 12/5/2012 | 4:41:00 PM
    re: 100G Watch: OFC/NFOEC Warmup

    Craig,  I believe your scenario for someone who has a fiber in Manhattan to probably be first for real deployment.  Those rings have been first for almost all new "higher speed" technologies to come out on optical transport (excepting reach) for over 20 years.  So, indeed, first single lambda, DWDM 100G dual polarization DQPSK with coherent reception will likely be on a link like this. 


    but will it be announced?  sometimes these are not.  Also same for certain key defense department links.


    sailboat

    wentriken 12/5/2012 | 4:41:03 PM
    re: 100G Watch: OFC/NFOEC Warmup

    @furious_george you do make the case that internet exchanges do have increased demands to serve.


    As far as 100GE having the opportunity to leak beyond peering sites, facebook and several other customers, I think this is limited by the fact that 1GE still a lot of capacity for customers, regular enterprises, and even moderate sized campuses.


    For short reach, the largest use I have heard for 100GE is to reduce fiber density and rack space -- never to increase capacity. This it-would-be-nice-if-it-was-available-and-cheap demand does not appear to me to support a healthy ecosystem. For long reach (>=10km), I have yet to imagine one potential customer for this technology. Only by looking at trends and letting my mind daze, can I imagine that the next data point will fit the curve.

    Page 1 / 3   >   >>
    HOME
    Sign In
    SEARCH
    CLOSE
    MORE
    CLOSE