[Federated-fs] Conference call today

[Ellard, Daniel]

If you didn't get the invite I sent out earlier, please let me know...

Some info from the IETF meeting on Monday:

If you were there and have additions/revisions/retractions, please mail
them to the list.  I presented an overview and status summary of the
fed-fs project (and since I was presenting, my notes were a bit spotty).

Lars Eggert, one of the area directors for the NFS WG, expressed
willingness to extend the charter and asked the WG chairs to propose an
extension that incorporates the fed-fs work.  The chairs asked whether
there was consensus that this work should be picked up, and from where I
was standing I saw all yes votes (zero nos, but a handful of
abstensions).  The next step is to prepare the requirements doc (which
we already have) to be added as a WG draft and help the chairs in
whatever way is necessary to make sure things keep moving forward.

Since we've already done a lot of work on the requirements doc, we're in
good shape on that.  We need to review it and see whether anyone has any
new requirements or whether we've drifted away from anything.  I expect
some new conversations to erupt as this work is taken more seriously by
more people.

One thing that came up at the IETF meeting (by Nico Williams from Sun)
is he thinks it's very important for the "discovery" of the root
server(s) to be automatic, and the mechanism should be a required part
of the protocol.  I think I agree.  This has implications for the way
the namespace may be projected differently to different clients,
however, so I expect this to generate some additional requirements and
perhaps another protocol (executed by the client) to do this discovery.
This could be a long and heated discussion, and so my suggestion for the
short term is to come up with something simple, practical and extensible
for version 1 and focus on not painting ourselves into a corner for
version 2.

As a meta comment, we could use a slide deck that people can use to
evangelize these ideas.  They're getting too complicated to explain in
20 minutes.

-Dan

-- 
Daniel Ellard, Ph.D.
NetApp Advanced Technology Group