Last Call Comments: draft-ietf-ldup-lcup-06

["Kurt D. Zeilenga" <Kurt@OpenLDAP.org>]

To the IESG,

It is our opinion that draft-ietf-ldup-lcup-06 (LCUP) should not
be published as a Proposed Standard.  We feel Experimental would
be more appropriate track.

The LDUP WG has been developing LCUP I-D for many years (based upon
prior individual work) but, to date, there is no "running code".
While "running code" is not a strict requirement of publication of
Proposed Standards, we believe this is one case where "running code"
is needed.  Content synchronization protocols can easily have
significant operational impact upon the Internet.  For example, a
single change to the directory could cause a large number of full
content exchanges to a large number of clients.  Such issues cannot
be adequately addressed without implementation and operational
experience.

We experimented with LCUP last summer.  We found that it was
infeasible to implement LCUP's incremental synchronization due to
is requirement for the server to maintain large amounts of detailed
history information.  LCUP does not provide adequate protocol
constructs to cope with insufficient history information.  If
we choose to support LCUP, we would have to force full reloads
for most (if not all) requests for incremental content update.  While
such an implementation could be viewed as conformant, such an
implementation would not meet our applications' requirements for
efficient and effective content synchronization.  Hence, we
abandoned our LCUP efforts.  We are currently pursuing alternatives
(which we intend to publish as Experimental once our I-D is updated
to reflect our running code).

During our time of LCUP experimentation, we found and subsequently
raised numerous of technical and editorial issues to the authors,
chairs, and/or WG.  Many of these issues have yet to be resolved.
It is our opinion that the LCUP I-D's quality is inadequate for
Standard Track approval.

We believe further experimentation in this area is needed and,
hence, suggest approving the document as Experimental (instead of
returning it to the LDUP WG to resolve known technical issues).

Kurt Zeilenga and Jong Choi