> And before anyone accuses me of sounding overly critical
> towards the AU ISP's, let me point out that we've dropped the
> ball in a major way here in the United States, as well.
We've dropped the ball in any place where the broadband architecture is
to backhaul IP packets from the site where DSL or cable lines are
concentrated, into an ISP's PoP. This means that P2P packets between
users at the same concentration site, are forced to trombone back and
forth over the same congested circuits.
This would seem to primarily be an issue /due/ to congestion of those
circuits. The current solution, as you suggest, is not ideal, but it
isn't necessarily clear that a "solution" to this will be better.
Let's look at an infrastructure that would be representative of what
often happens here in Milwaukee.
AT&T provides copper DSL wholesale services to an ISP. This means that
a packet goes from the residence to the local CO, where AT&T aggregates
over its network to a ATM circuit that winds up at an ISP POP. Then, to
get to a DSL customer with actual AT&T service, the packets go down to
Chicago, over transit to AT&T, and then back up to Milwaukee...
Getting the ISP to have equipment colocated at the point where DSL lines
are concentrated would certainly help for the case where packets where
transiting from one neighborhood customer of an ISP to another
neighborhood customer of an ISP, but in the common case, it isn't clear
to me that the payoff would be significant.
Getting all the ISP's to peer with each other at the DSL concentration
point would "solve" the problem, but again, the question is how
significant that payoff would be.
It would seem like a larger payoff to simply make sure sufficient
capacity existed to move packets as required, since this not only solves
the "local packet" problem you suggest, but the more general overall
problem that ISP's face.
And P2P is the main way to
reduce the overall load that video places on the Internet.