BBN Peering issues
Mike Gaddis
mikeg at savvis.com
Wed Aug 12 22:58:21 UTC 1998
Large access oriented companies like BBN, MCI and others have been keen to
point out the problem with the so called web-farms. In the early days it
clearly was a problem because the costs associated with carrying bandwidth
were not evenly matched between a large distributed (and costly) access
network and a MAE attached data center. It created an economic disparity
that could not be sustained. Let's recognize the validity of that argument.
Many of us (in the Brokered Private Peering initiative) have been looking
for ways to solve that disparity and to create a fair value proposition
between both types of networks. The best mechanism we have found is for
the web-centric companies to have regionally distributed connect points and
substantial backbone capacity to perform best-ingress or cold potato (from
web to access point of view) routing. In our view, this model reverses the
negative value proposition into highly favorable terms for the access
company. After all, their customers want access to these Web sites and the
service is now dropped off regionally, saving considerable backbone expense
for the access company.
That Exodus is doing cold potato and BBN does not find this sufficient to
consider Exodus a true peer causes me some concern. It makes one wonder if
there is a solution to this peering problem after all. I don't want to see
the government intercede without at least trying to find peering balance.
I believe Exodus has shown great willingness to "do the right thing here."
This can develop into a backbone civil war and I think BBN will not fair
well if that occurs. (There are major web sites being cut off if this
break in peering happens. Believe me, BBN customers will care a great
deal.) I don't see the upside in this decision. I truly hope that cooler
heads prevail, and soon, or major parts of the network may black out for
many thousands of Internet users. This will cause major harm to the
Internet market and the perception of any-to-any connectivity. This will
likely demand government intervention. Not good.
Mike Gaddis
CTO
Savvis Communications
-----Original Message-----
From: alex at nac.net [SMTP:alex at nac.net]
Sent: Wednesday, August 12, 1998 4:39 PM
To: Robert Bowman
Cc: dg at root.com; michaels at sun.aracnet.com; asr at millburn.net;
nanog at merit.edu; alex at nt.nac.net
Subject: Re: BBN Peering issues
On Wed, 12 Aug 1998, Robert Bowman wrote:
> I've tried to stay away from posting here, but a small clarification...
>
> Exodus is currently dealing with the asymmetry of traffic by assuming
> a large chunk of the burden by doing best-exit, at BBN's request, based
> upon BBN's MEDs. So let's not make it into the argument that the "evil"
> web hosters are just dumping recklessly via shortest-path out. Exodus
> has assumed the cost of all long haul delivery coast-to-coast of the
> data that BBN's users are requesting.
Heh, that makes it look even worse for BBN now and potentially in the
future, depending on what Exodus chooses to do. It's amazing that they
shot thierselves in the foot like this.
More information about the NANOG
mailing list