[c-nsp] IPv6 SLAAC on P2P or QinQ subints

[Mikael Abrahamsson]

On Tue, 6 Nov 2012, Tim Densmore wrote:

> Just a quick sanity check.  I'm in the early stages of planning customer IPv6 
> deployment and am trying to figure out how I'm going to deliver SLAAC over 
> P2P (ATM specifically) and QinQ interfaces for DSL or similar, in non-PPP 
> scenarios.
>
> When I started poking around, it became apparently quickly that "ipv6 
> unnumbered loopbackN" wouldn't work since loopbacks apparently can't/won't 
> send RAs, and this would break DAD in any event, or at least make is useless. 
> Digging around, the recommendations I've seen are to simply apply a static 
> /64 to each subint.  This seems absurdly cumbersome, even if using 
> general-prefix.  Am I missing something here or am I "stuck in ipv4 mode?" 
> Can anyone point me to a better way to handle this?

I think you need to elaborate what "this" is.

People typically use /126 for link networks, but you don't actually need 
them as a lot of things can be done using only link local. That however 
won't work with RAs (you don't specify why you want to use RAs).

So take one further step back and describe what you want to achieve.

-- 
Mikael Abrahamsson    email: swmike at swm.pp.se