Re: AS hopcount draft (redux)
Albert Tian <tian <at> redback.com>
2005-12-05 07:43:25 GMT
Yes I see this can be a simple and useful tool, and I support it to
become a WG draft.
Wondering would it be helpful if we allow the hopcount to be decremented
by a configurable "peering-distance" rather than a fixed number "one"
for ebgp. The default peering-distance can be zero for ibgp, one for
On Thu, 2005-11-17 at 03:25, Robert Raszuk wrote:
> > While I can see it seems simple, I don't see how it would be very
> > useful. As the mesh of interconnections gets more and more dense, I
> > don't believe it's really a sufficient mechanism to express (the
> > intended) policy.
> Your doubts that the idea may not address few possible interconnect
> topologies and may not be the universal solution for limiting update
> propagation for generations to come is IMHO not a sufficient argument
> not to progress with the draft and make it a WG doc to address all other
> much more controled AS meshes.
> Imagine there is a network operator who owns number of ASes and wants to
> limit some information distribution in one of BGP SAFIs just to his
> domain. The usual topology is one AS in the core surranded by reginal
> ASes. If he configured AS_HOPCOUNT to 3 and do not peer (by design) from
> the core AS his problem is easily solved.