Re: [Idr] WGLC on draft-ietf-idr-as-private-reservation-00 concluded, extended to consider ASN range

[Jeff Wheeler <[e-mail]>]

On Wed, Dec 19, 2012 at 9:57 AM, Jon Mitchell <[e-mail]> wrote:
> I agree 7% is too big.  2% does not seem unreasonable to me.

Do you foresee other kinds of large reservations being made from the
32-bit ASN space?

If NO, then 7% is not too big.  If it were, 2% would also be too big.
If you think public ASNs will be exhausted again in the future, and
having 5% more public ASNs will prevent that eventual exhaustion, I'd
love to see a basis for that prediction.

Basically, what I'm saying is, if 7% is too big then now is a good
time to start talking about 64-bit ASNs.  Everyone knew the 16-bit ASN
space would probably need expansion, and most folks I know came to
that realization in the mid- to late-1990s.  It's 2012 now and we
still do not have an extended BGP community formatted as 32:32 bits,
which is needed by numerous operators for BGP route advertisement
control.  Some major transit networks still don't have 32-bit ASN
support for their customers.  If 15 years isn't enough to go from
problem realization to feature-complete and widely-deployed fix, then
please, stop worrying about 7% or 2% and start crying that "the sky is
falling," so you can have 64-bit ASNs before you run out of 32-bit
ones.

If YES, you do imagine other large reservations being made from the
32-bit ASN space, have there been any proposals?  Will these
proposals, combined with existing public ASN use, exhaust 93% of the
32-bit space but not 98% of it?

The only "threat" I see to the 32-bit ASN space is the notion that RIR
policy should allow delegation of large blocks of ASNs to
organizations (companies, service providers) for whatever uses they
can imagine.  If that happens, I do not think +/- 5% global ASN pool
size will make a difference.  Either such RIR policy will run the ASNs
out, or it won't.

> Maybe some others would like to weigh in on one of the above options now
> that it's a focused discussion (otherwise I plan to make the small edits
> discussed in previous threads and move the start of range to
> 42xxxxxxxx so we can close this out...)

I asked some operators about decimal vs bit-aligned, and virtually all
feel strongly in favor of decimal-alignment.

I asked some to choose between starting a larger private range at
900k, 9000k, or 4.2b, and found there is much less interest in even
expressing an opinion at that point.

-- 
Jeff S Wheeler <[e-mail]>
Sr Network Operator  /  Innovative Network Concepts