Scaling ARIN proposal for small ISPs - and economic reality

[Stephen Satchell]

At 2:51 AM 1/21/97, David Hakala wrote:

>The average cost of an IP address drops precipitously under ARIN's initial
>proposal. For $2500 I can get 256 addresses or 8,192 addresses - a drop from
>about $9.76 each to under $0.31 each. The slope of the cost/address curve
>gets even worse as you get into the "medium and higher categories. This is
>stupid economics.
>
>Since public IP addresses are a finite resource, each additional address you
>want should cost *more* than the last one. That's how the real estate market
>works. As land becomes scarce, the price of a lot goes up. Why not license
>IP addresses one at a time for $10 each, 10 for $110, 50 for $1000, etc.?


If the sole concern was number space allocation, you'd be right.  What has
come out of much of this discussion is that there is a serious side issue:
backbone routing.  It's a technical thing, not political.  (At least not
political on the surface, although what technical discussions are ever
completely non-political.)  I'm still learning about the problems myself so
I can't talk intelligently myself.

The pricing structure is designed to encourage the consolidation of IP
addresses geographically (both in physical terms and in terms of the
connectivity topology) so to reduce the capacity requirement on the router.
You might think that routing is scalable, but some people I've talked to
(plus papers I have in my files from my APRAnet days) show that there are
significant problems that get real, real nasty as you increase the number
of nodes at a given level.

By the way, the gross revenue is to fund the registry and its operation.
The rate schedule is not intended to *increase* total revenue, but to use
revenue to encourage certain practices which are (arguably) in the best
interest of the Internet.

Stephen Satchell, a founding member of the Internet Press Guild