Re: There is no business case for broadband Internet [Was re: Interconnect Scaling]

[David Farber <dave () farber net>]

Begin forwarded message:

From: vijay gill <vgill () vijaygill com>
Date: April 15, 2007 2:54:16 AM EDT
To: dave () farber net
Cc: ip () v2 listbox com
Subject: Re: [IP] Re: There is no business case for broadband  
Internet [Was re: Interconnect Scaling]

Dr Farber, for IP if you wish. Just a few points to add here, so I've
snipped the vast majority of message.


> >
> >> Despite all the back and forth... it still isn't clear to me  
> whether
> >
> >> the Internet backbone, as most of the mainstream carriers / service
> >
> >> providers have structured their portions of it or are capable of
> >
> >> adapting, is capable of handling the vastly greater traffic load  
> that
> >
> >> the advent of P2P based HD video distribution networks will
> >
> >> precipitate. This seems to be a matter of:
> >
> >>
> >
> >> a) whether the interconnects between the various carrier / provider
> >
> >> network backbones are diversified and distributed enough to prevent
> >
> >> chokepoints from developing

This is a good point. However, upgrading the backbone paths are
trivial compared to upgrading the local metropolitan service areas.
things like hfc plant upgrades/node splitting and planning RF channel
capacity will dominate. It is a myth that p2p will reduce cost. It
will reduce cost for the folks distributing the data, but that cost
will just be shifted to the providers of the last mile. P2P shifts
costs around, it does nothing to reduce it OVERALL.


> >
> >> b) whether the architectural optimizations of your average P2P
> >
> >> network mean that vastly smaller amounts of traffic will  
> actually run
> >
> >> over the backbone and through any potential chokepoints than might
> >
> >> otherwise be the case... which also seems to be a dependent on what
> >
> >> percentage of the traffic generated will consist of widely popular
> >
> >> (and thus widely served) content versus specialized content that is
> >
> >> less likely to be available from a nearby peer (within a network
> >
> >> backbone boundary)

See above. P2P shifts costs.

> >
> >>
> >
> >> I suppose that, given the rapid and relatively painless (at least,
> >
> >> apparently so) emergence of Internet video over the past year  
> and the
> >
> >> (apparent) ease with which the additional traffic was accommodated,
> >
> >> we shouldn't worry too much... does anyone have year over year
> >
> >> figures on the total volume of Internet traffic and the percentage
> >
> >> increase year over year? I tried hunting this information down in
> >
> >> Google, but wasn't successful....

ATDN has published graphs for their growth, and  Level 3 mentions
traffic carried on their backbone in their filings.



http://www.renesys.com/blog/2006/03/a_tale_of_four_carriers_att_ve.shtml
has some interesting points.

Also

Level 3 continues to experience rapid customer growth on its network.
The Level 3 global IP backbone network currently carries over 3.7
petabytes of Internet Protocol (IP) traffic every day (one petabyte is
the equivalent of 2,000,000 CDs). IP traffic carried across Level 3's
transatlantic network has doubled in the last twelve months.

http://www.level3.com/newsroom/pressreleases/2006/20060209.html


> >
> >>
> >
> >> Regards,
> >
> >> Thomas Leavitt
> >
> >>
> >
> >>
> >
> >> Dewayne Hendricks wrote:
> >
> >> [Note:  This item comes from reader Mike O'Dell.  DLH]
> >
> >>
> >
> >> From: mo () ccr org (Mike O'Dell)
> >
> >> Date: December 12, 2006 9:23:38 AM PST
> >
> >> To: dewayne () warpspeed com
> >
> >> Subject: interconnect scaling
> >
> >>
> >
> >>
> >
> >> the mystical powers of "large peering points" still lives, i see
> >
> >>
> >
> >> for the better part of a decade, large ISPs have used multiple
> >
> >> point-to-point interconnects between their networks, situated
> >
> >> by the demands of network engineering, not the location of
> >
> >> "peering points".
> >

Actually there are several points that are the "new" MAEs and the
constraint on those are no longer bandwdith to a switch, but rather
availability of power and space. We are constrained on expansion by
lack of power in most facilities.



> >>
> >
> >> the *only* magic in MAE-EAST (or MAE-WEST) was that once upon
> >
> >> a time, when the total traffic to exchange was a couple of T1s
> >
> >> worth with 4-5 other ISPs, there was an economy of scale in
> >
> >> buying those T1s as a cross-town DS3 (or later, metro ethernet.)
> >
> >> everybody got their DS3 to the same place and a switch was put
> >
> >> there to connect the dots.  once the traffic between pairs of
> >
> >> ISPs got big enough, the economic optimization that was MAE-EAST
> >
> >> and MAE-WEST fell apart.
> >
> >>
> >
> >> that's all they ever were - a policy-neutral economic optimization.
> >
> >>
> >
> >> and one that went deeply sub-optimal many, many years ago
> >
> >> (at least for large ISPs - they still get used by smaller
> >
> >> players whose traffic exchange needs can still take advantage
> >
> >> of that particular economy of scale.)
> >
> >>
> >
> >> -mo
> >


/vijay


-------------------------------------------
Archives: http://v2.listbox.com/member/archive/247/=now
RSS Feed: http://v2.listbox.com/member/archive/rss/247/
Powered by Listbox: http://www.listbox.com