Policy Statement on Address Space Allocations

Wed Jan 31 19:09:39 UTC 1996

    From: Vadim Antonov <avg at sprint.net>

    BTW, "settlements for routes" is a rather interesting thing. Imagine,
    say, Sprint peering with Joe Blow Internet-And-Burgers.  Let's assume that
    route costs $100/yr, so Sprint pays $1M/yr to JBIB and JBIB pays back
    $300/yr.  Sounds like a good business for JBIB :)

That why I've always thought that *iff* we have route charging (not sure it's
a practical solution, but I think about it), the charge should be based on the
*scope* of the advertisement. A route that has to be advertised globally thus
costs more.

In this case, "Sprint" would pay JBIB a routing fee based on the size of JBIB
times the number of routes Sprint injects. JBIB's fee would be the sum of {
for each route JBIB injects, the sum of the routing fees to carry that route
everywhere it goes in the 'Net }. (Presumably Sprint will have a charge based
on the size of its net, etc, etc.) On average, if JBIB is injecting a small
number of routes, this work out close to a wash (small no of routes time large
scale being about equal to large numebr of routes over small scope)

This is an ideal model, in practise it could well be too complex, but it's
useful to think about.

    The better idea is auctioning and trading the routing slots, though it is
    unclear where the proceeds from the initial auction should go

This assume that all slots are global in scope. This is unlikely to be the
case. My suggestion was that each ISP auction all the slots in its routers,
the money going to the ISP. This produces (in practise) the system above,
where the cost of a route is related to the scope over which it is advertised.

This has a number of interesting long-term stabilization properties. If the
number of routes grows too fast (i.e faster than technology) it will cause the
cost of routes to go up, producing back-pressure to rearrange addresses so
that we can do the same job with less routes. Also, if a key ISP doesn't
provide enough routing slots, or gouges people for them, in either case
driving up the price, that will increase the cost of using that ISP, and a
competitor can snarf the business.

    I don't know what you make of it, but i think that sucks a lot more than
    the zero-sum model we have now.

Well, it's certainly more complex, and maybe unworkably complex.

On the other hand, if there is a problem, *something* that can handle it will
evolve one way or the other. If community pressure on people to be good about
routes won't do it, life suggests that some other means, a combination of
legal and commercial (since that is how we allocate scarce resources in this
world, which is what routing slots will be), will arrive.

	Noel