Comments welcome.
note: AD hat on.
Bob & Margaret have sent me a revised IPv6 charter that I have
forwarded to the IESG/IAB for consideration. It is largely consistent
with the charter that the WG has already seen and supported by the WG
with one exception. "DNS discovery" is not included in the charter.
I asked that this item be removed from the charter for several
reasons:
- If I were to take it to the IESG/IAB (who get to review all WG
charters), I would expect to get strong pushback on including the
DNS discovery work in the IPv6 WG. Reasons would include:
- this is not IPv6-specific work, it should be done in a broader
IETF context, e.g., where the DNS operational expertise resides
such as in DNSOP.
- why not just use DHC? I know that many are opposed to including
DHC as part of the solution, but that view is not universally
shared, understood, or agreed too. Especially outside of the IPv6
community.
- what about the technical concerns that have been raised regarding
the approach in the current draft? (See Erik Nordmark's
presentation at the Yokohama meeting.) IMO, there has been a lack
of rigorous consideration of the issues that have been raised.
Although I have heard the arguments coming from proponents of the
DNS discovery work, I have not been strongly convinced. Moreover, I
also have heard the arguments against the approach in the current
document from others, and find their arguments to have merit as
well. That puts me in the awkward position of arguing for a position
within the IESG that the WG supports, but that I have concerns with
and where I don't have strong answers for addressing those
concerns. The issues here are not black-and-white. I see that there
are multiple sides to the argument, but this group seems to have
difficulty in having a technical discussion about those issues. It
seems like the WG is to a large extent tired of this issue and just
wants to have an RFC.
- the dns discovery effort has dragged on for way too long (more than
2 years), and due in part part to the history, I have me doubts
about this WG being able to have a productive discussion on the
topic anymore. Being very aware of ongoing discussions and concerns
being discussed on, e.g., the problem-statement list, it is time to
make some decisions and move on. One of those decision may just be
that the idea of a quick-and-simple DNS discovery has missed its
window of opportunity, and we just need to move on.
- the current approach of interest assumes the use of SL
addresses. Given the issues swirling around SL and to what degree
the WG wants to encourage or limit their usage, I find it hard to
believe that the current ID is a good way to go forward. It would
seem odd to recommend this approach while simultaneously trying to
discourage the usage of site-local addressing.
- it is not clear to me there is really consensus for the current
document at this point in time, whatever past concensus there might
have been. Moreover, it's not clear how much support the selected
approach has outside of the IPv6 community. This gets back to an
earlier point about doing the work in consultation with the right
experts.
Where to go from here? If there is truly interest in continuing this
work, it might make sense to engage the DNSOP WG to see what kind of
interest there is there _for the problem statement_ before even
discussing any particular approach. It might also make sense to host a
BOF on either the narrow topic of DNS discovery or the broader topic
of service discovery. Steve Deering posted a long note a while back on
"serverless discovery" that I think was quite interesting, but much
work needs to be done to flesh that out, assuming there is interest in
doing so. Finally, there is always DHC (the IESG approved DHCPv6 as a
PS at the end of 2002), so the IETF _does_ have a recommended approach
-- it's just one that some people don't like.
Thomas