RE: comments on draft-durand-v6ops-assisted-tunneling-requirement s-00 .txt

["Karim El-Malki (HF/EAB)" <karim.el-malki@ericsson.com>]

Hi Marc

>Karim,
> "dumb questions":
>- NAT were not planned to happen. IPv4 was not designed with NAT. NAT
>happened. Nobody can control the deployment of NATs.
>- now, when you say: 
> + "the customer does not have a NAT in the customer equipment", how can
>we
>enforce/ensure that this will never happen?

Well, now there is a deployable alternative with IPv6.
It's impossible to be 100% sure of things, but saying that it
is OK to put a NAT in a mobile phone at the time when mobile
manufacturers are going to dual-stack terminals doesn't make
sense to me.
  
>
>It appears to me that NATs are everywhere, even in places that tried to
>avoid them.(The world's richest organisation in terms of IPv4 addresses
>have so many NATs you can't think of...). And you can't restrict the
>deployment of NAT: the operator of a network (3G, wireless, wired,
>etc...)
>may deploy a NAT as its own way. How would 3G be able to enforce the
>non-existence of NATs?

There are no NATs in mobile phones today since there is no use
for them and I sure hope never to see them. I think most people
would agree. So there is nothing today that hints at this
development and I think we should not be proposing solutions
that are compatible with this.

Mobile manufacturers will be shipping dual-stack mobile terminals
and that is what we should be (and are) favouring. If a PAN is
needed behind the terminal then it's an excellent case for using
IPv6. Supporting NAT traversal mechanisms in this scenario seems
very dangerous to me and it doesn't help since we really want to
reduce the use of NATs if possible, not increase the traffic
through them.

>Does the assumption of no-NAT really stand?

I think it does.

/Karim