Re: draft-huitema-v6ops-teredo-03.txt

["EricLKlein" <ericlklein@softhome.net>]

From Alain

> > I am still a little uncomfortable about the fact that there is no
> > explicit
> > statement that "teredo is needed because NAT was deprecated in IPv6
>
> I have never seen any such statement in any RFC defining IPv6.
>
The very conecpt of unique addressable addresses negates NAT. As it is there
will be problems created as people try to use the 10 net addresses fro
mtheir IPv4 network on the IPv6 network as those addresses going to pop up
in lots of places. Also, there was a large discussion on depreciating NAT
last year. I was against depreciating as too many people would resist. This
was why draft-vandevelde-v6ops-nap-01_draft1.txt was written.

>
> >  so
> > teredo is needed to allow IPv6 nodes to communicate with IPv4 nodes in
> > a
> > domain that utilizes NAT."
> >
> > This is still true in section 3.2.4 Automatic sunset.
> > Some how this whole section scares me as it seems to imply that there
> > is a
> > way to turn an IPv4 NAT into an IPv6 Router while maintaining the NAT
> > like
> > functionality. As stated above, NAT is not part of IPv6 (NAP is but is
> > still
> > not mentioned in this document).
>
> 3 points:
>
> a) nothing will ever prevent anyone from inventing/using IPv6 NAT one
> day
>
You are correct, explicit RFCs stating that NAT is not permitted will have a
limited effect, and as you said these have not been published. RFC 3879
Deprecating Site Local Addresses  explains why private addresses (like the
10 net) are not good and why they should not be used.

> b) when the v4 NAT gets upgraded to do also v6, it still does v4 NAT...
>       this functionality does not go away...

That is correct, the concept goes away. Unique addressable addresses and NAT
are mutually exclusive concepts.

> c) The only (minor) point I have with section 3.2.4 is with the
> sentence:
> "upgrading the
>     Internet connection used by the NAT to a native IPv6 service,"
>
> If the 'upgraded' NAT provides v6 connectivity via a configured tunnel
> (maybe using the tunnel set up protocol we want to design in v6tc)
> teredo will also detect it (by seeing a native RA on the internal
> network)
> and turn itself off.
>
> So I would suggest to simply remove the work 'native' from the above
> sentence.

The point is that "native IPv4" vs "native IPv6" was added because of the
working group's decision to do away with NAT. We had a defined "private"
address space in the original IPv6 RFCs, but this was removed in RFC 3879
Deprecating Site Local Addresses. Given this the WG has been working towards
preventing IPv6 NAT.

Eric