On 22-mrt-2007, at 19:02, Dino Farinacci wrote:
As I said, BGP and DNS can both be considered to be mapping protocols, and they both have some aspects that are desirable and some aspects that are problematic.
After this week, I have concluded that both BGP and DNS *should not* be considered. However, we could model something new off of BGP and DNS. That is, a pull versus push model or a combination of both.
Indeed.
BGP on the other hand floods everything but has serious scaling issues and suffers from the problem that local optimizations introduce global state.
BGP does not flood everything. On eBGP peering it sends only best paths.
(iBGP does that, too.) I mean everything in the sense of all prefixes.
What we need in a new mapping system is the scalability and parallelism of the DNS coupled with the immediacy of BGP, but where BGP distributes next hop information, the new mapping system should distribute "last hop" information.
What is last-hop information?
In BGP, there is a "next hop" attribute that pretty much does what the name suggests, we'd expect nothing less from a routing protocol. However, in an id/loc (or loc/loc) mapping system, we're not coming dealing with the next hop the packet is sent to, but rather, the place where it can be decapsulated, which can be considered the "last hop" although that's probably not 100% accurate. The intermediate hops don't interest us here, either we assume that if a locator address is present in the mapping database it's reachable (and we flood unreachability information) or we put in an additional protocol that checks whether the decapsulation box is reachable.
_______________________________________________ Int-area mailing list [email protected] https://www1.ietf.org/mailman/listinfo/int-area
