On Mon, Jan 12, 2009 at 12:34 PM, Tony Li <[email protected]> wrote: > > Hi Danny, > > |> I know that the internal tables are always a pain, but since > |we have > |> to deal > |> with the global issues, the internal, private growth (self- > |> inflicted ;-) has > |> to necessarily be out of scope for RRG. > | > |That's quite odd to me, considering by today's definitions > |"internal [BGP] tables" are where the routing scalability and > |stability are it's worst - today, and unquestionably, the first > |place things will break IF/WHEN they do, as a result of an > |inter-domain routing protocol architecture that forces either > |full-mesh or hierarchies such as route reflection that themselves > |introduce additional paths and state in the network (even with > |implicit aggregation effects). > > > As you point out, this is more to do with the specific protocol than it is > to do with the overall routing architecture. As such, this is why it's > really much more of the domain of IDR than RRG. >
So, it's not that RRG doesn't acknowledge this problem, it's that the problem ought to be fixed (to some extent at least) with the work RRG is doing? > |And they're not going to break because of a 100k unique > |internal-only routes, they're going to break because of > |an order or magnitude or more paths (and all of their overhead) > |- paths introduced as a result of "global issues" and solutions > |that focus on solely minimizing DFZ size, rather than looking > |at where the problem is actually worst - today. > > > Paths that are introduced for the sake of traffic engineering are a well > understood and self-inflicted problem. If those folks that introduced the I don't think it's just 'traffic engineering' that is the problem here, it's often other things like L3vpn/2547-vpn/mpls-vpn, 'internal services' or 'service networks' (Peter's notes about phones on IP roaming around the sprint core). Surely TE prefixes are issues, but that's not the only source of internal route growth... one large one is 'this IP thing, it's kinda successful!' (convergence). -Chris (btw, I support the idea that RRG's work will 'fix this problem', my original note was just to continue to raise awareness that the numbers for route growth need to keep in mind that 285k prefixes globally is often dwarfed by internal table sizes.) _______________________________________________ rrg mailing list [email protected] http://www.irtf.org/mailman/listinfo/rrg
