a question about the economics of peering

Bill Woodcock woody at zocalo.net
Fri Nov 30 18:28:21 UTC 2001


      Alex Rubenstein wrote:
    > Today, I was approached by *unnamed-ethernet-extension-company*.

I take it this was PacketExchange?  There are several other companies in,
or entering, that market right now as well.  Most of them haven't a clue
about the economics of the field they're in, and will go down in flames as
quickly as they can burn their first-round funding.

    > My belief is that generally speaking, networks meet at public
    > exchange points to exchange traffic with each other more
    > economically.

Yes, you're correct.  Specifically, one participates at exchanges that
offer sufficiently attractive routes such that one's per-bit delivery cost
is, on average, lowered as a consequence of participating at the
exchange.  That is, the total cost of participation in the exchange must
be less than the cost of delivering the same number of bits via transit.
That's why many of the most successful exchanges are also often the
cheapest, with the fewest services...  They exceed that price/performance
threshold more easily than exchanges that feature expensive services.  Um,
that's something of a tautology, sorry, but I think you get my point.

    > His belief is that people will pay a premium to get to an exchange
    > point

I hope he believes in Santa Claus too, else his company is going to
encounter a rude shock when it tries to make a sale.

    > Essentially, he said that paying more for peering that for transit is
    > typical, and to be expected, and most people accept this.

Nope.  He's as wrong as can be.

                                -Bill






More information about the NANOG mailing list