Re: v4 NAT vs NAT-PT models

[Suresh Satapati <satapati@cisco.com>]

> In turn, I'd request you to stop spreading the gross simplification
> that NAT-PT is equal to plain old v4 NAT, and NAT-PT works as
> specified without DNS-ALG.  This is definitely not the case.

When I referred to the model, I meant with NAT or NAT-PT, for certain
applications (think non-DNS) you need the binding to be setup during
signalling. This could be done by an ALG (or your favorite) residing on
the same or a different box that does header translation. This would enable
subsequent data/media to go through using the bindings. Maybe I wasn't clear
about this earlier.

>
> If we refer someone to RFC2766 for translation, there are two options:
> either implementing DNS-ALG or inventing a replacement on your own; I
> don't think there exists even an internet-draft describing other
> possibilities to achieve the same effect.  Making such a referral
> without explicit mention to requirements regarding DNS-ALG or similar
> behaviour is necessary to avoid causing any more confusion.
>
> Of course, vendors are free to deploy any unspecified mechanisms they
> want.  But that doesn't fix the problem for those who haven't
> developed an unspecified mechanism for achieving the same effect;
> for all intents and purposes, unless something has been specified or
> documented, it does not exist.

You have combined two arguments, which is actually misleading.

model(that i mentioned above) + DNS_ALG being a part of NAT-PT ?
We can keep arguing about this and get no where. This will be my last mail
on this topic.

>
> > I'd like to get a sense on where the WG Chairs stand regarding:
> [...]
>
> I take it that you're asking for an "official" standing.  Could you
> clarify what the question is, or would you just want us to comment?
> I'll add that on our agenda, but the response could take a while..

Yep. official standing. That's what I meant.