Re: a comment on nat requirements draft

[Jari Arkko <jari.arkko@piuha.net>]

Marcelo,

> what people had in mind was that the NAT box could perform the RO on 
> behalf of the v4 CN, but some people already said that this wasn't 
> that useful neither, what do you think?

Uh-oh.

That might be doable in the base RFC 3775 mobility, but not in any of 
the extensions (shared secret keying for RO, for instance).

I question the goal to attempt full feature parity. I would spend your 
cycles more wisely and try to make sure that if both v4 and v6 are 
available, you choose v6, making it possible to use any feature.

Or in my opinion this is right at least for mobility route optimization. 
It is indeed just an optimization. Even with direct IPv6 connectivity 
there is no way to guarantee that the optimization is supported by the 
implementation of the peer, accepted policy-wise by the peer, or passes 
through firewalls in between. That's why the optimization is designed to 
be fail-safe. If you can turn it on, great. If you cannot, things will 
still work.

Jari