RE: spending time on analysis [Re: draft-palet-v6ops-proto41-nat-03 as WG item]

["Christian Huitema" <huitema@windows.microsoft.com>]

> On Fri, 14 Nov 2003, Fred Templin wrote:
> > Then, let's have ISATAP as a WG item; it's been around for three
> > years and there is nothing premature about that. (Better yet would
> > be to just go ahead and publish it NOW.)
> 
> We don't yet know whether we need ISATAP.

Correction. We very well know we do. In fact, we know it so much that
there is actual deployment at many sites, and interoperable
implementations by several vendors. You can only write that "we don't
know" if you define knowledge in the very narrow sense of "we do not
have a requirement document that explains why we absolutely cannot do
without it."

Let's go back to April 2002, when the IESG decided to close NGTRANS and
to put a temporary stop to the definition of transition mechanisms. The
feeling at the time was that having too many mechanisms would create
confusion, and that we should only develop those mechanisms that
correspond to a clear scenario. 

That sounded logical, but the implementation of that idea in NGTRANS
turned out to be terrible. The WG embarked in an exercise of defining a
set of transition scenarios for various deployment areas, and the
exercise proved both contentious and cumbersome. It took us about a year
to agree on the first scenarios (3GPP and unmanaged), and we have yet to
complete the ISP and enterprise scenario. The obvious explanation is
that the world is very diverse, and trying to have a single document
reflect everybody's ideas of a transition requires lengthy negotiations.
The negotiations continue with the requirement documents, which may be
even more contentious.

The example of ISATAP is quite telling. It would be very easy to write a
simple 2 paragraph document explaining how ISATAP can be used to provide
IPv6 connectivity in an enterprise network as an overlay over the
existing IPv4 infrastructure. We may not unanimously approve that this
is needed, but we should never require unanimity in such cases. If there
is a substantial constituency that approves the requirement, the IETF
tradition is to acknowledge that the requirement exists.

The V6OPS working group is a prime example of the current IETF decease:
expanding a lot of energy to achieve no result. The symptom is very
clear: vendors are shipping products based on internet drafts, without
getting the benefits of IETF peer review. 

I would propose one simple way to meet the "describe your scenario"
requirement: ask anyone who proposes a transition technology to include
in the draft a two paragraph description of what scenario the proposal
is supposed to facilitate. The WG should then verify that there is some
demand for the scenario, but should not require unanimity or even "rough
consensus": just because someone believes I don't need something does
not change my need for it. The role of the working group should be peer
review: make sure that the proposal is well engineered and meets it
stated requirement without causing harm.

We must break the current analysis/paralysis.

-- Christian Huitema