Follow-up work on NAT-PT, 10 October 2007
|Follow-up work on NAT-PT, 10 October 2007
|Metadata last updated
|Send notices to
To: The Internet Engineering Steering Group
Subject: Follow-up work on NAT-PT
From: "Olaf M. Kolkman"
Date: Wed, 10 Oct 2007 21:34:10 +0200
Cc: IAB IAB , IETF at ietf.org
The IETF has previously standardized NAT-PT as a means of
intercommunication by packet translation at the Internet layer. However
subsequent analysis revealed a number of issues with NAT-PT that make it
a poor choice for a general purpose interconnection mechanism. The IESG
recently approved RFC 4966 which states:
Although the [RFC2766] form of packet translation is not generally
applicable, it is likely that in some circumstances a node that can
only support IPv4 will need to communicate with a node that can only
support IPv6; this needs a translation mechanism of some kind.
Although this may be better carried out by an application-level proxy
or transport-layer translator, there may still be scenarios in which
a revised, possibly restricted version of NAT-PT can be a suitable
solution; accordingly, this document recommends that the IETF should
reclassify RFC 2766 from Proposed Standard to Historic status to
avoid it from being used in inappropriate scenarios while any
replacement is developed.
As noted in the approved text quoted above, there may still be scenarios
in which a revised, possibly restricted version of NAT-PT may be
For example, one such scenario may be increasingly likely given the
current rate of consumption of IPv4 address space. Namely, there may be
legacy IPv4-only hosts in one part of the Internet that want to talk to
dual-stack hosts in another part of the Internet that do not have public
IPv4 addresses. The servers follow today's recommendation to be dual-
stack, but the scenario still does not work, therefore the general
recommendation that servers be dual-stack is clearly not sufficient.
Furthermore, classic v4-v4 NAT is also not sufficient for this scenario.
For example, one cannot put multiple web servers behind the same classic
v4-v4 NAT since web browsing uses port 80, and the NAT has fewer public
IP addresses than the number of servers behind it or else the NAT
wouldn't be needed.
While the IETF has reclassified RFC 2766 as Historic, up to now there
has been no replacement to RFC 2766 for such scenarios. The IAB note
that the IESG has now begun discussion on what work is needed for
transitioning to IPv6 and where it should be done, and the IAB
encourages the IESG to strongly consider the scenario outlined above
when chartering IETF work.
-- The IAB