>> Michel Py wrote: >> I will point out though that these issues are not limited to a system >> where routers are rewriting packets and that it is likely going to be >> easier to tackle them by containing the translation into a smaller >> number of devices (routers) vs. doing the the id/loc mapping in each and >> every host.
> Keith Moore wrote: > I have serious doubts about that; to me the boundary between API > and host-to-network interface seems much cleaner and more definite > than the boundary between one side of a router and another. That's only half the picture; host people tend to like host solutions and router people tend to like router solutions. You missed that half of the picture with NAT: yes NAT sucks, it does not change the fact that nowadays there are more NAT hosts than non-NAT hosts. And you also missed that half of the picture by pissing so many vendors about the NAT issue that made a point of not implementing 6to4 in their NAT boxes just to piss you off in return, where having 6to4 implemented in NAT boxes would have been by far the best promotion avenue for 6to4. Don't try to sell me that crap, I'm not buying it. Anyway, you missed the point: Although there are things (such as NAT) that eventually are worthy adopting the ideology-only "just say no" line (if done right), you can't apply this to everything and in the case of where the id/loc occurs we are looking more at a "whatever works" situation. Ideology is one thing, first try to overcome the handicap that id/loc mapping at the host level will generate hundreds or thousands more _times_ requests than on a router-based basis, two or three orders magnitude more. Given what you have posted recently, I guess it's not going to be DNS based. When you have solved the issue, come back and argue the argument again. Michel. -------------------------------------------------------------------- IETF IPv6 working group mailing list [EMAIL PROTECTED] Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
