> *If* we want to pursue this path at a minimum the host needs to know > whether using the original 1st hop router is better or not. > One way to do this is to require that routers on the same link > must advertise the same set of prefixes (and make the router > advertisement > consistency in RFC 2461 check this?). > If we did that then the hosts could assume that when > different routers > advertise different sets, the routers are not managed by the same > entity and the host should use a 1st hop router that is > consistent with > the source address. But the host could switch 1st hop > routers among the > set of routers that advertise the same set of prefixes.
=> I think that it's generally a good idea to recommend that, despite the fact that we don't yet have a full solution. I.e. there is nothing to lose if we recommend that and there are potential gains. This will also simplify MIP movement detection a bit. Should we strengthen the language on this in 2461bis? There is already a recommendation that routers should check other RAs for consistency, but I don't think it says anything about advertising the same prefixes. Perhaps we can add that as a "RECOMMEND". > > I don't know whether we could make the operational > requirement that the routers > on the same link (that don't want source-based routing) must > advertise the > same set of prefixes stick. Depends on what the current > operational practices > I guess. => Agreed. But then again better now, when there is not global deployment, than few years down the road when it is widely deployed. Hesham -------------------------------------------------------------------- IETF IPv6 working group mailing list [EMAIL PROTECTED] Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
