Generation of traffic in "settled" peering arrangement

Vadim Antonov avg at
Wed Aug 26 19:31:44 UTC 1998

Sean M. Doran wrote:
> Remember Yakov's route push and route pull model.

These are pretty much equivalent - if both parties follow
them and have similar mix of packet sources and sinks.

My take is that the simplest combination of "neutral
ground" public IXPs and no-transit backbones contains an
interesting economical negative feedback effectively
preventing market monopolization.  (I.e. the bigger
backbones have to carry more traffic, thus neutralizing
advantages of the economies of scale).

In this respect, the mandated settlement-free NAP
connectivity (no matter how disliked by backbone ISPs) is
a good thing. Eventually FCC will have to come up with
interconnection rules, and i suspect they will take a dim
view on the continuing backbone consolidation and
exclusion of smaller players.

OTOH, the costs of peering with a lot of small folks are
non-trivial.  I would expect IXP operators to start
offering the route clearing-house services on commercial

Assuming that the mandated settlement-free exchanges are
likely to be the future direction of regulatory push, the
hot-potato routing seems to be the best approach.  You
simply act in your selfish interest by pushing packets off
your backbone asap.


More information about the NANOG mailing list