[ppml]How far is too far?

[Geoff Huston]

At 11:00 AM 4/10/2003 -0400, Marshall Eubanks wrote:
>On Sat, 4 Oct 2003 10:06:54 -0400
>  "McBurnett, Jim" <jmcburnett at msmgmt.com> wrote:
> > > You need to be careful here.  AS numbers are actually a more limited
> > > resource than IPv4 addresses!  While I haven't looked at the
> > > ARIN graphs
> > > recently, a couple years ago ASN exhaustion was the most immediate
> > > threat, although plans were being discussed to extend it....
> > > --
> >
> > Please forgive my ignorance here.
> > But after the dot-bomb shouldn't a large number of ASN's
> > now be unused?  and not just that, what about the
> > mergers? uunet alternet worldcom etc...
> > When looking at the netatlantis.org site, I seem to
> > see large gaps..
>
>
>15776 in my BGP address tables, 30,000 or so assigned or allocated (IANA
>isn't talking to me this morning...).
>
>Exhaustion looks to be about one doubling away, with a doubling time
>of ~ 5 years.
>
>So you might be able to recover 50% if you were very aggressive, which 
>would buy
>you 3 years or so, which I regard as highly optimistic.


I did some work on this earlier this year, and presented it at the March 
IEPG meeting
prior to the IETF.

http://www.potaroo.net/iepg/march-2003/BGP%20AS%20Number%20Exhaustion.pdf

At current rates the 16 bit AS numbers will exhaust somewhere between 2009
(no reclamation efforts) and 2011 (complete reclamation)

I don't think we have deviated from this projection over the intervening 
months.

It would appear to be prudent that we all swing into using 32 bit AS numbers
somewhere around 2006, in which case extensive AS reclamation would be an
exercise of dubious benefit.


Geoff