[nfsv4] IETF 107 session dealing with new work

[David Noveck <[e-mail]>]

Of the two sessions requested for IETF 107, one is to be devoted to future
work.  The other deals with the current wg agenda and was described in mail
I sent out 12/17.  The future work session will primarily consist of
discussions of proposed new work items for the wg. but I'm also going to
carve out a portion (esrtimated  at 40 minutes) discussing possible
improvements in how we do things.  More about this later

As long as all proposals for new work are within our charter, as I expect
them to be, we should be open to a wide range of proposals and arrive at an
agenda in late February driven primarily by what people are willing to
present and what people want to hear about and discuss.

For now, proposals for talks  or discussion topics should be sent to the wg
list.  If the proposals start looking like they will fill up the time
available or more, the priorities should be based on what attendees,
including remote attendees, want to hear about.   I anticipate that, as we
move into February, the discussion will start focusing on the relative
priority of the various proposed items, if the time is oversubscribed.   If
there looks like there will be free time, we can discuss how any available
time would best be used.

So far, in the new work category, Ithe following proposals have been made,
totalling 50 minutes:

   - Chuck is anticipating spending 10 minutes on the poor performace of
   directory operations and what we might do within the protocol to enable
   improvements.   There is no current I-D for this.
   - Chuck is also anticipating discussing requirements for computational
   storage, based on his document draft-cel-nfsv4-comp-stor-reqs.
   - I anticipate spending 15 minutes discussing parallelization of
   metadata operations.   There had been some work in this area but nothing
   really came of it, most likely because it was focused on providing
   within-single-directory parallelism which requires a high degree of degree
   of agreement between clients and servers as to directory organization, in
   order to support striping.  I'd like to discuss our options regarding the
   simpler (one hopes) task of providing within-file-system parallelism while
   keeping the handling of each directory on a single server node.   There is
   no I-D for this but if there is sufficient interest, I will write one after
   the meeting.
   - I'm allocating 15 minutes for a discussion of Sorin's compression
   attributes proposal, based on the then-current  iteration of
   draft-faibish-nfsv4-data-reduction-attributes.

As far as the ways-of-working sub-agenda, there are two things currently on
the list.

   - I've asked Chuck to lead a discussion of the state of the document
   review process and what should be done to improve it. I've allocated ten
   minutes  based on Chuck's time estimate.
   - Chuck will be giving a talk about using github for managing personal
   drafts and Working Group documents.   He intended to give this talk at IETF
   105 but there was not enough time.  I've allocated ten minutes.

That accounts for 20 of the 40 minutes reserved for this area.   If there
are further proposals in this vein, let me know by 2/15.