[ppml] Policy Proposal 2005-9: 4-Byte AS Number

Michael.Dillon at btradianz.com Michael.Dillon at btradianz.com
Thu Dec 22 08:53:05 EST 2005

> Conversely, there is such a thing a two bytes of AS number.  They are
> not represented as ASCII strings on the wire.  The fact that you see
> them as decimal numbers is simply an artifact of where you sit.  For
> instance, see rfc1771, section 4.2 and look at the diagram of the open
> message.

I see that as an implementation issue, not a policy issue.

> IF it is postulated that migrating to a longer ASN representation over
> time is a good thing, AND it is understood that this is experimental
> at this time, AND it is understood as a consequence of previous
> experience that having registry services from day 0 is a Good Thing,
> THEN it makes perfect sense to have a policy that says "registries
> will make this available".  It's unclear to me from your email whether
> you have a problem with this.

I agree that the registries should make available
AS numbers greater than 65535 for those organizations
who need to put the new BGP protocol into production
and who need to ensure that the quirks of large AS 
numbers (mapping to AS 23456) are working OK.

> I have a lesser degree of support for the sunset provisions that Geoff
> includes in his proposal, and I think he and I are going to have to
> agree to disagree on whether this should actually be three separate
> policies.

Since ARIN has revised the policy set into a policy 
handbook, I don't think it matters how many changed
clauses are included in a policy proposal. However,
the proposal does need to be clear, understandable, 
and easy to relate to the existing policy set. I would
prefer to see plain English wording, and I would prefer
to see the proposal as a list of adds, changes, and
deletions to the existing policy set.

In one respect, you are right that not all elements
of Geoff's proposal are equally urgent. So in the 
interests of simplicity, it makes sense to deal
with first things first.

--Michael Dillon

More information about the ARIN-PPML mailing list