On Thu, Aug 15, 2002 at 12:54:39PM -0700, Chris Wedgwood wrote:
On Thu, Aug 15, 2002 at 10:04:49AM -0700, Joe Abley wrote:
If there's ever a chance that an advertisement needs to be advertised as multiple long prefixes instead of (or as well as) a covering aggregate for reasons of inter-domain traffic engineering/site multihoming/whatever, then the filters in place need to be ready to accommodate that.
It seems unlikely that this is the case for all of these networks, especially /21s smashed into /24s.
It's extremely common for edge networks to pre-arrange a capability to deaggregate in case they ever need to. Some large operators build deliberately permissive filters to accommodate this, without even being asked; the direct risk to the transit provider from rapid deaggregation can be mitigated by using prefix filters together with maximum prefix limits. Multi-homing by poking PA holes in aggregates and advertising the results in different directions can also require liberal filters at the transit provider boundary. None of these things necessarily correspond to instantaneous observed policy in the network at any particular time, anyway; at best, they represent hints to peers about what the future might hold, possibly, maybe. So the RIB would be a better place than the filter to point a finger at if you wanted to complain about route aggregation, but even then, unless you're standing close to the origin of the routes in question there's every chance that your criticism is misdirected. The argument that deaggregation is bad and wrong because it hurts the core is interesting right up until the point the customer decides to buy service somewhere else because you won't let them multi-home the way they want to. Joe - To unsubscribe from nznog, send email to majordomo(a)list.waikato.ac.nz where the body of your message reads: unsubscribe nznog