Re: [sig-policy] prop--065: Format for delegation and recording of 4-byt
Geoff Huston said the following on 20/8/08 14:57:
>
> Because you and Gaurab have both decided to propagate this rather silly
> notion in the last day or two,
Please don't shoot the messenger.
There is wide belief in the operations industry that the RIRs have
decided to adopt a dot notation without paying any attention to what
people who run networks actually have to face. The belief has come about
as operators are now closer to the reality that they actually will have
to support customers appearing with 4-byte ASNs in their networks, and
have started looking at traffic engineering configurations, provisioning
systems, flow analysis tools, etc.
> and because the "they" you refer to
> appears to include myself in my day job, I should respond to correct it.
>
> The original policy proposal regarding the 4 byte AS numbers put to the
> RIRs had a section called "Nomenclature" and proposed a colon separator,
> which, as far as I recall, I picked up from original notes by Enke Chen
> in the context of the IDR working group of the IETF when they were
> developing the work.
>
> Noone from any region make a single comment on the nomenclature section
> of the proposal at the time, with the exception of the ARIN community,
> where the general response after some considered discussion on the
> matter was the view that a colon would make parsing BGP communities
> tricky, so why not use a period in place of a colon as a delimiter of
> the low and high bytes of the AS value. Subsequent review of the policy
> proposal in ARIN generated the comment that nomenclature on the dot
> notation was out of scope for policy, so why not drop this section
> completely. So it was dropped from the policy proposal not only in ARIN,
> but, as this was a coordinated policy across all the RIRs, it was
> dropped from the proposal for all the other RIRs as well.
Yup, this tallies with my recollection too. But it's not what many
people are saying, so I'm happy that you have taken the time to clarify,
thank you!
Was very aware that the colon caused confusion concerns; the dot was the
following proposal, but I don't recall any broad industry support for
that either. There were various incarnations of
draft-michaelson-4byte-as-representation-xx.txt which did document the
"dot", but that series of IDs does not seem to have proceeded anywhere
(please correct me if I'm wrong).
> So as far as I can see where we are with AS notation largely as a result
> of consulting with, and listening to, the community.
The nomenclature piece was out deemed out of scope at ARIN. But the RIRs
seemingly ignored it and carried on using dots anyway. Who decided? ;-)
Thanks for confirming my point.
So IANA use the dot in documenting their allocations to RIRs, the RIRs
use dots in their documentation... And implementers are using all three
notations, or not, and it's all quite confusing for the end operator.
As we are getting closer and closer to the date where the RIRs will be
handing out 4-byte ASNs by default, more and more operators are looking
at what they'll have to do to support them. I've highlighted a couple of
things that will have to be reworked within ISP operations; I've not
pointed out the work needed to revise provisioning systems or netflow
analysis tools to use anything apart from the normal integer format that
has been good enough for the last 15+ years.
As a postscript, I'd like to point out to everyone on the list the so
far positive support for draft-huston-as-representation-00.txt
(stipulating 'asplain' representation for 4-byte ASNs) on the IETF IDR
Working Group mailing list. Geoff, I'd like to wish you and George best
of success with it!
Best wishes,
philip
--