RE: GCI and PEER Network


Subject: RE: GCI and PEER Network
From: Leif Sawyer (lsawyer@gci.com)
Date: Tue Dec 18 2001 - 13:16:10 AKST


Jim Courtney writes:
>
> Not only that, but ISP's have to buy more bandwidth on the
> fiber when their across-town packets are forced down to Seattle
> and back. This is a bad situation (for consumers, not GCI) that
> the marketplace has not yet addressed. Right now it's mostly game
> servers - later it will be local phone service that will take a
> beating with latency if you're not in the
> 'big-digital-mondo-cable-bundle' plan.
>
> If you have most of the subscribers, you win. One day someone else
> will figure out a way to capitalize on this situation and even
> things out.

I think that the ultimate fate lies in the future of IP telephony.
If ISP's embrace this as a way to side-step LEC's and LD carriers, then
they'll probably start to look at peering interconnects to decrease
their upstream bandwidth.

Of course, it's going to take a lot of monitoring and data massaging
to even figure out what what the statistical breakdown of traffic is,
in order to justify local NAPs, and what sort of business model you
would need in order to pay for them.

Who pays more?

If ISP A, which has tons of IPtelephony traffic flowing over to ISP B,
but ISP B has minimal originating IPtelephony traffic terminating on ISP A,
it's not in B's interest to support the NAP. They'll want A to pay for it.
"Hey we noticed a ton of IPt traffic terminating from you. It's probably
 chewing up your backbone. Y'all should buy a [T1] of interconnect from US
 to offload from your upstream providors!"

Not that this model isn't valid today -- it's just harder to convince them
of it because web-surfing is all entertainment. It's not toll-bypass
telephony.

Leif
(again, my opinions, not my employers)



This archive was generated by hypermail 2a23 : Tue Dec 18 2001 - 13:16:22 AKST