RE: Comments on draft-ietf-multi6-v4-multihoming-02

[Geoff Huston <gih@apnic.net>]

> I'd propose that sending keep-alives might be a tunable setting, and the 
> timing
> of sending keep-alives could be set according to policy (host or site).


If you are talking about what makes multi-homing trigger a location change, 
then the observation as I see it is that the existing amount of information 
passed vertically in the protocol stack is inadequate. Wether its 
keepalives, transport level timeouts, ICMP messages or any other means of 
detection of 'failure' you need to be able to pass this information to the 
point where a multi-homing homing change can be undertaken. I'd agree with 
John's perspective that it is not easy to pre-determine the settings that 
relate to failure detection, and a reasonable response is to make such 
settings tunable.

  Geoff