Remi, >> injecting IPv4 routing into the IPv6 routing table is considered a bad idea. >> that was what caused Automatic tunnelling and IPv4 compatible address in >> RFC1933 to be deprecated. > > - Reference?
good question. anyone aware of an IETF document, giving a detailed description of why we deprecated automatic tunnels? the best I can come up with, regarding IPv4 routes in IPv6 is: http://www.windows-ipv6.org/index.php?page=implementation&selectlanguage=eng > - Should we take it, then, that you reject moving from DS routing to > IPV6-only routing without address/prefix changes as a use case you are > interested in? *talking about the general case here* IPv4 addresses are used to number the outside of a CPE NAT, and can have relatively short address leases today. the resulting renumbering event only affects ongoing TCP sessions. given that, it doesn't sound like the main use case for MAP. it can be implemented in MAP, with a single rule as well; e.g. by assigning a separate IPv6 prefix /128 prefix to each CE. what are we arguing over? if 4rd-U is better than MAP? in the MAP effort we chose to drop quite a number of suggested features, because we didn't think the added complexity/confusion was worth it. the working group are free to add any feature from the feature buffet. cheers, Ole _______________________________________________ Softwires mailing list [email protected] https://www.ietf.org/mailman/listinfo/softwires
