Paul A Vixie <paul@vix.com> wrote:
I would like to know if anyone has measured this one way or the other, since if there is a demonstrated tendancy toward local traffic, it may open some currently-closed minds on the value of joining *hundreds* of regional IXPs and regionalizing our routes so that we can inject a subset into each such IXP without giving anyone unintended transit or subsidizing their long haul costs.
What do you propose to do with route flap? While creative "regional" filtering is theoretically possible i'm convinced the complexity involved will quickly overshadow whatever benefits the local peering provides. Long-distance telcos generally find it cheaper to backhaul traffic hundreds of miles to large switch sites instead of installing zillions of small switches. And, no, public exchange points are not dead. They have benefits of higher flexibility (one 155M pipe going to an exchange point is _more_ than three 45M pipes going to three other ISPs -- the traffic is not spread equally, and there are fluctuations). There is at least one way to make public exchanges to handle a lot (10000 times) more traffic. But you know it already. --vadim