On Mar 29, 2011, at 3:32 PM, "Alain Durand" <adur...@juniper.net> wrote:
>> The routing optimization only works if is no IPv4 address overlap. But of >> course there will be IPv4 address overlap. Here is how the routing >> optimization outright fails -- Alice wants to communicate with Bob. But she >> will instead be route-optimized to Carol: This is a transition mechanism, not a full solution. If Alice wants to talk to Bob, she should consider using IPv6. Indeed, her inability to communicate with Bob over IPv4 could easily be described as a feature. Furthermore, if Alice and Bob were both behind NATs, communication between them would be similarly problematic. A fairly thoroughly-realized kludge does exist which works in some cases, but it definitely breaks under at least one CGN deployment with which I have painful experience. The bottom line is that transition mechanisms are not, and should not be, held to the same standards as native internet protocols; if it were, we would have to make the transition to IPv6 through a flag day. _______________________________________________ Int-area mailing list Int-area@ietf.org https://www.ietf.org/mailman/listinfo/int-area