On 10/07/2009, at 1:07 PM, Curtis Villamizar wrote:
In message <7fa8c7b6-ace4-4604-98ec-2585b937b...@apnic.net>
Geoff Huston writes:
The way I've seen this phrased is that ASes X, Y and Z announce an
aggregatable set of prefixes to AS A, and AS A wants to to announce a
single covering aggregate.
Geoff
That is certainly the simplest case. Difficult to handle variations
are an A and B or A is one hop removed.
The concern for the simple case is that A isn't authoritative for the
prefixes that A is aggregating. The issue with the more difficult
cases is that the boundary of aggregation needs to be defined somehow
such that providers know which should be announcing the more specifics
and how far they are supposed to go and who is authorized to announce
and aggregate containing those more specifics.
I'm not at a provider anymore so I don't know how badly aggregation
practices have deteriorated or what the registries are up to.
You are right that there may be indirection going on here and a more
general description is that ASes X, Y and Z announce an aggregeatable
set of prefixes and AS A wishes to announce a single covering aggregate.
The difficulty here lies in attempting to describe the authorities
that may need to be involved here but of course in the scope of the
use cases document such difficulties can be conveniently ignored
Geoff
_______________________________________________
sidr mailing list
sidr@ietf.org
https://www.ietf.org/mailman/listinfo/sidr
