|From:||"CAUCHIE Gregory RD-CORE-ISS"|
|Date:||Thu, 26 Jul 2007 16:09:09 +0200|
as opinions were requested tuesday during the rtgwg meeting regarding interests on IGP muti-topology applicability, here is mine.
Today, most of multicast services are deployed using the PIM protocol. With PIM, multicast trees are set up by relying on IGP informations, thus decisions to send join and prune messages are made upon an IGP adjacencies status basis. But for multiple reasons, the IGP adjacencies states may not exactly overlap the PIM adjacencies states. In that case, the shortest path from one router to a multicast source or a RP can include a link where at least one PIM interface is down. The consequence is that PIM join messages are sent on a dead end and then multicast service is broken until some other actions (that may request human direct actions) are made.
To prevent this issue to occur and thus unavailability of multicast services (like IPTV), multi-topology is a real interesting solution. Multi-topology could be use to convey in LSPs or LSAs information of links where a PIM adjacency is UP. Thus LSDB would then have a "PIM topology" in parallel of a classical network IGP topology so that the former could be use as a basis for SPF computation from the router to the multicast source instead of the latter. The PIM join and prune messages would then be sent on the best "multicast available" way if it exists.
This is to me the first interest of multi-topology work.
_______________________________________________ rtgwg mailing list [email protected] https://www1.ietf.org/mailman/listinfo/rtgwg
|<Prev in Thread]||Current Thread||[Next in Thread>|
|Previous by Date:||rtgwg meeting at IETF-69, John G. Scudder|
|Next by Date:||Draft rtgwg minutes for July 24, 2007 meeting, John G. Scudder|
|Previous by Thread:||rtgwg meeting at IETF-69, John G. Scudder|
|Next by Thread:||Draft rtgwg minutes for July 24, 2007 meeting, John G. Scudder|
|Indexes:||[Date] [Thread] [Top] [All Lists]|