The persistent questions of what "locator" and "identifier" mean came up on the LISP WG list. I didn't think it was useful to continue there since it's for working with particular protocols, not theory. But Noel and I continued on the side. I said I wished someone else could benefit from the discussion. Noel suggested I post something I said to him here, so he could reply. Here it is, somewhat edited down.
----- Excerpts from Scott Brim <[email protected]> ----- Date: Fri, 13 Mar 2009 16:12:43 -0400 From: Scott Brim <[email protected]> Subject: Re: [lisp] [ipdir] LISP WG To: Noel Chiappa <[email protected]> Excerpts from Noel Chiappa on Fri, Mar 13, 2009 02:12:21PM -0400: > > From: Margaret Wasserman <[email protected]> > > > I am concerned about the accuracy of calling this mechanism an > > ID/Locator split mechanism > > Well, if it is not intended to separate location and identity, > what's the point of creating a mapping database, to maintain maps > from one namespace to another? It maps between routing scopes. In both cases you are dealing with things that name network attachment points. However, even though they all name network attachment points they are not _routed_ the same. So the mapping system answers the question "how do I get this packet to a place where the routers will have enough information to forward it onward?". > And two of the capabilities it explicitly provides/supports are > provider independence and multi-homing - two things for which > separation of location and identity are generally held to be fairly > crucial. They provide provider independence at the site level, not endpoint. The endpoint still does not have any kind of topology-independent identifier it can use if it, itself, is multihomed or mobile. (abstraction: ... it needs an identifier if it uses multiple interfaces in close proximity for the same session, where "proximity" means either in space or time) > If one has to pick one _short_ phrase to describe the goals and > point of LISP, it is indeed 'separate location and identity'. 'reduce strain on the routing system by separating site routing from core routing' > Only when one gets to a longer description does one have the room to > add the complex caveats of 'not complete separation _in the initial > stages_, because initially LISP EIDs still retain some location > semantics, and also name interfaces as opposed to stacks'. > > Yes, it is not absolute separation of location and identity (at > least in the short term), but that is because LISP made the > strategic choice to support _unmodified hosts and site-local > routers_ Right. However, over the months we've realized that in a world where MOST devices will be wireless and many will have multiple radios, we will have multihoming and mobility in the endpoints, so this isn't enough. That is not to say there is no reason to separate sites from core. We need both. - Sites need protection at their core edges to protect them from having to renumber. That way they can change providers at will. This can be done by map-and-encap, NAT, something like ILNP, or Virtual Aggregation. - Endpoints need to be able to use multiple attachments and change their attachment points at will. They need identifiers that support that. They can use either higher layer identifiers (e.g. Trilogy multipath) or lower layer ones (MIP, HIP) that decouple the usual tuple from actual addresses. ----- End excerpts ----- _______________________________________________ rrg mailing list [email protected] http://www.irtf.org/mailman/listinfo/rrg
