Why is this scalability problem ? A trivial answer: because we do user reachability dissemination ( everyone may add: but otherwise there were no internet). Non-trivial answer:
Because we still depend on user reachability dissemination ! Loc/Id-split is the right way to look at the issue. But by providing also “location” similar to postal letters where both the receiver AND its location are written on the envelope, we should make sure to get the same benefit, i.e. the elimination of the scalability problem. LISP however does not abolish the dissemination of user reachability info, it only replaces it by RLOC reachability dissemination combined with EID-to-RLOC mapping dissemination. Just old wine in new hoses. TARA however will enable forwarding to any destination router without looking at the receiver’s IP address, i.e. doesn’t need the dissemination of user reachability info at all. I am thinking about a strategy such that TARA could be incrementally deployed so that the BGP-Table will incrementally shrink until it becomes obsolete. Note, China transformed from communism to capitalism. The revolutionary was, that it didn’t take a revolution. Likewise BGP can be transformed from a non-scaling, topology-agnostic protocol to a forever scaling, topology-aware protocol (there is no law which says only link-state protocols can yield a topological graph :- )! Heiner **************
_______________________________________________ rrg mailing list [email protected] https://www.irtf.org/mailman/listinfo/rrg
