I have heard several arguments for why people might like to use some sort of limited-scope addressing in special circumstances. I've also heard a couple of general benefits (long-lived connections and dubious security benefits).
But, I haven't really heard any arguments that have swayed me from my original position. Private addressing causes significant complexities, and we don't know how to solve all of the problems that it causes for applications, DNS, routing protocols and network management protocols. Therefore, I don't think that we should define it as an inherent part of IPv6. There may be some significant advantages to private addressing, so I _do_ support the idea of an IRTF WG being formed to sort out the issues surrounding private addressing and make recommendations regarding what the IETF should/shouldn't standardize in this area. I also think that an IAB document in this area would be appropriate. But, I don't want to see us standardize a badly-thought-out private addressing scheme at the core of IPv6. In my opinion, there are three possible choices here: (1) Limit the scope of the IPv6 WG's problem, by forbidding the use of site-local addresses on globally-connected networks in the scoped addressing architecture. This would not preclude work in this area by other groups within the IETF/IRTF, and we could remove the restriction when the implications are fully understood. (2) Develop a complete solution for scoped unicast addressing within the IPv6 WG. This would include solving the problems they cause for all protocols/layers. (3) Define an IP-level solution for scoped addressing (similar to what is currently in the scoped addressing architecture), and consider all of the implications of that architecture on other protocols/layers to be someone else's problem. The first choice is both responsible and doable. I have deep concerns about the second choice, along two lines: (1) I think it is important that we stabilize and complete IPv6 quickly, as folks are widely implementing and deploying it, and (2) I don't know that we have the expertise (in the IPv6 WG or anywhere in the IETF, without further research) to solve these problems. And, in my opinion, the third choice (which is what we seem to be doing so far) is blatantly irresponsible. Are there people who want to argue for choices #2 and/or #3? Or are there other choices that I've left out? Margaret -------------------------------------------------------------------- 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] --------------------------------------------------------------------
