> I'm having a lot of trouble understanding what we are trying to accomplish > with this email discussion. From my, probably simplistic, view we are > missing a fundamental construct of IPv6 addresses. That construct is that > v6 address have some kind of scope attached to them. That scope is meant > to limit how far they can reach or where they can be used. Am I > understanding the v6 address architecture correctly?
Perhaps. But what we're finding is that while scopes might be necessary for corner cases (e.g. LL is certainly useful for configuration purposes), it's a bad idea to expect everything (hosts, apps, routing) to have to deal with them. And there's also a limit to which it's a good idea to wire notions of scope into the addressing architecture, and reserve even more address blocks for scopes that have dubious utility for the future. Just as wiring in notions of "class" into IPv4 turned out to be a limitation, wiring in more notions of "scope" into IPv6 might also turn out to be a limitation. > In between these 2 extremes (LL and Global) there is a range of other > scopes. Margaret pointed out some of these including different divisions > in a single organization or manufacture/supplier relationships. The Site > Local (SL) address was suppose to enable these types of relationships. That's not my understanding of the intent of SLs, and at any rate SLs as currently specified don't work well for this case - there are too many constraints and assumptions attached to them. > The talk of GUPI addresses doesn't seem to capture this limited scope > function. Am I missing something? Limited scope is a bad idea for anything that's going to be used outside of an isolated environment. > We certainly have a good handle on what a global address should do, and the > LL addresses are also well defined. The trouble is that we never had SL's > before and the concept of a 'site' is ambiguous so its hard to figure out > what to do with them. It's worse than that. a- there are *conflicting* assumptions about what to do with them. b- SLs are a major pain for applications that have to span scope bounaries. But let's not rehash that discussion again. It's in the list archives. Keith -------------------------------------------------------------------- IETF IPng Working Group Mailing List IPng Home Page: http://playground.sun.com/ipng FTP archive: ftp://playground.sun.com/pub/ipng Direct all administrative requests to [EMAIL PROTECTED] --------------------------------------------------------------------
