On 3 Jan 2013, at 22:05, Jared Mauch wrote: > Oh, I understand all these use-cases, but there is a case for a well-designed > network not always sharing/mixing the NLRI. eg: We don't transport v4 NLRI > in v6 transport, nor v6 NLRI in v4 transport. If you have a single session > with massive shared fate, perhaps it's not a protocol design error but a > network design error.
I would highlight that public Internet networks are NOT the only implementation of BGP. Yes, in the Internet you can divide the two topologies that you have (the v4 DFZ and the v6 DFZ) into separate sessions to separate fate. However, a (I will assert, significant) number of operators run networks which run one address family that has numerous entirely separate topologies within it (e.g., L3VPN). There are no mechanisms that provide separation between such routing topologies within that AFI, SAFI, and it is not practical or scalable to divide these into individual sessions per-topology. One of the taxes that we have with MP-BGP is that there are multiple applications which may not all have exactly the same deployment characteristics, or sensitivities. IMHO, we need to make sure we are considering all applications of BGP when considering this subject (this is one of the points that the draft that this thread relates to tries to cover). Having said that, where there are means where failure domains can be limited e.g. multi-session and/or separating AFI,SAFIs onto different transport sessions, of course, these form part of the solution. Cheers, r. _______________________________________________ GROW mailing list [email protected] https://www.ietf.org/mailman/listinfo/grow
