On Wed, Oct 03, 2012 at 09:36:58PM -0400, Russ White wrote: > All that matters is that I draw the same traffic into my AS, where the > more specifics still exist, so I can optimally route the traffic out of > my AS --the same as I always have. You seem to be thinking that this > would be done at the edge in some way --that the originating AS would be > doing the suppression. That's not the way this is designed to work at all. > > The suppression is supposed to take place at least one hop away from the > route origin, and only if I can be reasonably certain that my > suppression of the longer route isn't going to change traffic patterns > from my perspective.
AS 1 has 10.1.0.0/16 AS 2 has walked away with 10.1.1.0/24 AS 1 doesn't providing routing for AS2's prefix. - This means that any traffic for AS 2 via AS 1 is dropped. - This also means that AS 1 gets traffic for AS 2 in some cases, especially during events where AS 2's routes are dropped. They were aware of this when they signed the contract to let AS 2 use that address space. AS X receives the /16 and the /24 from upstream. AS X bounds the /24 and only advertises the /16. AS Y receives only the /16 from upstream. AS Z is dual-homed to AS X and Y. AS Z receives the /16 from AS X and AS Y. If AS Z chooses AS Y's /16, it has no reachability to AS 2. If AS Y chooses AS X, everything works. If AS X didn't implicitly aggregate the route, AS Z could use either AS X or Y. Summary: In cases where a subnet doesn't provide full reachability to contained more specifics, suppressing the more specific may result in lost reachability downstream. a. This is less likely to be the case when the origin ASes are identical. b. Even so, this doesn't preclude partitioned ASes. Providers may leak the more specific for an internal network when internal topology isn't sufficiently diverse to cover certain types of outages. See use cases for accepting one's own AS in an AS_PATH (junos 'loops', similar knobs in other implementations). c. You'd be correct that even in cases like b, when you're "far enough away", implicit aggregation is probably safe again. Such arguments have been used for route 'hop-limit' proposals. The usual counter argument usually devolves down to Internet core-meshiness making hop-limits somewhat useless. -- Jeff _______________________________________________ GROW mailing list [email protected] https://www.ietf.org/mailman/listinfo/grow
