RE: Consensus on identifier/locator split?

["Christian Huitema" <huitema@windows.microsoft.com>]

> Actually, if you reuse MIPv6 mechanisms to handle the "existing TCP
> connections" case, you might be able to get both levels of functionality
> without any changes.

That is certainly a reasonable option. In fact, back in 1995, the reason we did not pursue the "TCP optimization" path was that the "binding update" mechanism of MIPv6 was thought to provide equivalent functionality. MIPv6 also provides a reasonable "opt-in" mechanism: hosts implement the correspondent node facility if they want to enable TCP connection to survive a "re-homing event" at their peers; applications can opt for binding to a "long duration" source address if they want to use long duration connections, and to the "short term best" if they don't, or if they have another way to deal with mobility.

Which is to say, I would be much more confortable examining a specific proposition, such as "use of MIPv6 to survive re-homing events", rather than a generic proposition of "let's split".

-- Christian Huitema