Jonathan,
On Fri, Feb 15, 2008 at 05:49:35PM -0500, Jonathan Rosenberg wrote:
A big mistake was made in IPv4, where NAT was declared 'evil' and we
didn't spend enough time defining it well. Now, it is wildly successful
and a part of what the Internet is, and it is harder to deal with it.
Had we done standards work up front and early, and defining exactly how
NAT work, things would work much better. We should have had RFC4787 in
1997 and NOT 2007.
NATs are *not* on the wire protocols but middleboxes that break havoc
with peer-to-peer applications but that help to get people that don't
have enough IP addresses to use or have reasons why they cannot do a
network renumbering. For interoperability, there is no necessity for
all NATs to work in exactly the same way, hence the incentive for
anybody to follow a standard would be rather low. If we had defined
NAT in 1997, it would have been obsoleted before it had even reached
the RFC editor as competition in the marketplace would have forced
vendors/open source community to leap frog each other with small and
big improvements over the IETF standard. The only useful role for IETF
would have been and still is to provide some beHAVIORal advice on what
we have observed as a common lowest denominator between the different
implentations.
There is nothing special about NATs, we know what problems they cause,
we know what problems they solve. We even have relatively simple
protocols that can traverse them. Observations that it is hard to
deploy new transport protocols are not exactly very new either and it
is quite obvious that NATs are part of the story why deploying anything
new on the Internet has become much harder. Can we perhaps move on to
a topic that involves new insights or ideas ?
David Kessens
---
_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
http://www.ietf.org/mailman/listinfo/ietf