Re: [Idr] WG Last Call on draft-ietf-idr-as4bytes-11.txt

In message <6.2.0.14.2.20051013143340.02e25900@localhost>
Geoff Huston writes:
>  
>  
> > > It's the other way around more likely. The edges of the internet are
> > > more likely to go NEW first...
> >
> >    We've agreed that we should spec a transition which can work
> >either way (backbone going 4-byte first or edges going 4-byte first).
>  
> I do not understand this. Surprisingly, he earth is not flat.

That is only a theory according to some proposed science curriculum
revisions in the US.  [Is that too US Centric a joke?]

> What is
> an 'edge' and what is a 'backbone' are from an individual network
> operator's perspective often arbitrary distinctions, and then talking
> about which is these "goes first" is perhaps unhelpful What I like
> about the draft as it stands is that it views transition as a set of
> OLD / NEW and NEW / OLD transitions. That's simple and effective.  I
> would not readily ascribe attributes to a transition spec that starts
> from a premise of "backbones" and "edges" and then starts to talk
> about which "goes first".
>  
> regards,
>  
>      Geoff

Edges are things that are single homed or at most dual homed but the
important thing is that they don't take full routes, don't provide any
transit, and don't readvertise any routes that they learn from one
Internet facing interface to another.

Would you prefer transit and stub to backbone and edge?

It doesn't matter if single homed stubs ever understand as4byte.  They
don't really need to run an external routing protocol.  The dual homed
non-transit might take a subset of routes, but if a few are missing
they'll never know since they have a default (other than load balance
might be affected).  As soon as an AS passes routes from one peer to
another it might start to matter, but many small providers don't
provide transit to any BGP peers and their customers are not (or
should not be) passing routes to their other providers if dual homed
to more than one provider.  These "edges" don't have to convert before
a transition occurs.

It doesn't matter which AS converts to as4byte capable first as long
as the deployment is very far along before 4 byte AS start showing up.

Curtis

_______________________________________________
Idr mailing list
Idr@ietf.org
https://www1.ietf.org/mailman/listinfo/idr