Le 26 oct. 2010 à 16:47, Margaret Wasserman a écrit : > Remi, I have read at least two versions of the SAM draft, and I have talked > to you about it at length.
Yes you took time to have explanations about draft-despres-sam-02 in July 2009, telling me that it was to report to Jari. Following that, I don't remember having seen any comment or questions from you on it, or on its successors. > I've also talked to Mark Townsley about it, in an attempt to understand it. > I certainly don't remember every detail, but I do basically understand it. What follows tells me that you only "believe" to have understood it. > Unless it has fundamentally changed since the last time I read it , SAM > requires changes at every site This sentence shows that you don't understand that each site can become SAM-capable independently of others (like sites supporting 6rd of RFC 5969). This hasn't changed. > and, ultimately, on every host. This shows that you don't understand that all hosts that want to take advantage of PA-based multihoming with SHIM6 will have to do something new anyway. (Which doesn't mean that all hosts have to want it.) > Deploying something like that is on the scale of deploying IPv6 -- it would > take 10 or more years to do it. I am aware that you attribute a lot of > benefits to SAM, but for the purpose of people trying to bring up IPv6 > networks today, SAM doesn't even exist. > I have not ignored you. I have spent hours trying to understand your > specification, and I just don't agree that it is a superior, or even > workable, solution to this problem right now. It depends on what you consider to be "this problem". In any case this shows that you don't understand that SAM does work for SHIM6 to be usable in multihomed sites having PA prefixes. It also suggests that you feel secure that NAT66 won't create new problems with IPv6-capable hosts hosts as they are today (presumably using IPv6 addresses according to RFC 3484, at a time when a revision of this RFC is under study, to be deployed in all IPv6-capable hosts). I would not be so confident. > ... > Nothing that requires hosts to know and use their global addresses is > compatible with the notion of Address Independence. This confirms that you don't understand that, in a site that routes ULAs, SAM-capable hosts can use their global addresses, which they know, in e2e packets tunneled across the site. > Dead stop. Your choice. Fortunately, Dan and Jari with whom I chatted in Maastricht have been more encouraging. The next draft on SAM for TRAnsparent Multihoming (SAM/TRAM), which I couldn't have in time for this meeting, should be better in many respects. I hope you will then be more open to accept that it does useful things that aren't done otherwise so far. RD _______________________________________________ nat66 mailing list [email protected] https://www.ietf.org/mailman/listinfo/nat66
