Re: IPv6 Multi-homing BOF at NANOG 35

[Geoff Huston <gih@apnic.net>]

> Doesn't this apply to any orgization that doesn't have its own /32?


In the fullness of time this is the expectation, yes. However its pretty clear
that the approach entails some changes to the protocol stack, and its not going
to be a completely straightforward task. So its a pretty likely outcome 
that the
'conventional' practice of advertising more specifics into the routing system
will not disappear overnight.


> Doesn't this apply to any orgization that has customers that require
> multihoming but lack their own /32 as John points out?


as above.

> Doesn't this apply to any orgization that has it's own IPv6 aggregate,
> and wants to advertise something more specific than its aggregate?
>
> If that is the case, then it won't just apply to ISPs or multi-homed end
> sites who can't get a /32, it also applies to ISPs and end sites who have
> a /32, but need to split route announcement to share load across multiple
> transit providers.


I can appreciate the logic here- but in the same way that it strikes
me that using routing to perform fine-grained traffic engineering is akin
to using a mallet and an axe for a task that normally requires a micro-surgery,
this also strikes me as a very round-about way to get to a useful approach
to realize localized traffic engineering.


regards,

    Geoff