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

On 10-nov-04, at 19:03, Geoff Huston wrote:

>> IMHO: In a modular multi6 design, this shouldn't be prescribed.
>> The connectivity detection component might use keep-alives on 
>> Tuesdays,
>> SNMP traps on Wednesdays, and hints from the ULP the rest of the week.
>> This really should not become an architectural choice.

> precisely - and I believe that this was noted in the architecture draft

I think it makes more sense to accept hints from elsewhere, but do a 
quick additional check inside the multihoming protocol too. This way, 
SNMP traps (?) and ULPs don't have to bother implementing all the nitty 
gritty security details but they can just hand off their information 
and be done with it.