Final LDUP WG Meeting Minutes

See text below.

Chris Apple - Principal Architect

DSI Consulting, Inc.

mailto:[email protected]

http://www.dsi-consulting.com

=========================================================================

LDAP Duplication/Replication/Update Protocols WG (ldup)

Tuesday, March 18 at 1300-1400

===============================

CHAIRS:	Chris Apple <[email protected]>
	John Strassner <[email protected]> 

Minutes taken by:  John Strassner

The meeting was run according to the posted agenda. The
meeting minutes therefore mirror the agenda topics.

1. Consensus on a path to conclude - Chris Apple reporting

It was obvious from the last meeting that we weren’t going
to be able to achieve consensus to produce standards-based
documents for all of the work that is listed in our current
charter. After an email exchange, consensus was reached that
the working group would try and move all documents forward
on an experimental path. The one exception to this is the
LCUP document,  which is targeted at the standards track.

There were no comments to this presentation.

2. Status of various documents.

2a. LDAP Client Update Protocol - Rich Megginson reporting

The URL for this is:

http://www.ietf.org/internet-drafts/draft-ietf-ldup-lcup-04.txt

Rich opined that the draft is good as is and ready to be
released, with the one exception that Kurt Zeilenga and
John Young are not happy with the draft. Kurt has posted
a counter draft, called LDAP Content Sync, available
at the following URL:

http://www.ietf.org/internet-drafts/draft-zeilenga-ldup-sync-01.txt

First, it should be noted that the two drafts have moved
closer to agreement. However, there are still significant
differences between these two drafts, and the nature of
the disputes are fundamental to the design of each. The
disputes boil down to three points of contention.

The first point of contention is that while both protocols
synchronize directory information, LCUP operates within
the user information model, while LDAP Content Sync
operates within the system information model. An example
of a facsimileTelephoneNumber implemented as a virtual
or a collective attribute was used. The difference between
how the two protocols operate for this example is two-fold:

LCUP provides the changed data, LDAP Content Sync doesn’t.

LCUP causes every entry with the collective attribute
definition to be reported as modified (which could in
turn cause a large number of entries to be returned to
the client), while LDAP Content Sync only returns one entry.
The difference in the above is that with LCUP, the client
doesn’t need to know the details of the physical information
model, whereas with LDAP Content Sync, the client must know
the details of the physical information model.
 
The second area of difference is that LCUP requires
history to be maintained, while LDAP Content Sync doesn’t.

LCUP sends one entry because it has historical data present;
LDAP Content Sync sends the full requested entry if modified,
or sends the DN or UUID of the entry for unchanged entries.
This means that the client is responsible for deleting
entries from its local store that are not returned. This
in turn means that if there are N entries in the result set
of the search, modifying one value will cause N messages
to be sent.

The final major difference is that in LCUP, changes to meta
data, as well as subtree operations may either cause the
client to resync, or to send a large  number of messages.
This is because LCUP assumes that these operations are
very infrequent and therefore relatively inexpensive.
LDAP Content Sync doesn’t have this side-effect because
it always sends a message for every entry.

Kurt stated that as designed, LCUP doesn’t meet the needs
of OpenLDAP. Rich stated that the issues that Kurt is
raising are beyond the original design scope of LCUP.

The co-chairs recommended two action items, which were
accepted by the meeting attendees:

1) Rich and Kurt should meet and try and resolve their
differences by COB Friday this week. If their differences
can’t be resolved, then they should report that to the
working group.

2) Both drafts will need applicability statements added
if they want to be progressed.

The co-chairs will propose a recommended course of
action based on the results of the meeting of Rich
and Kurt.

2b. LDUP Update Reconciliation Procedures - Steven Legg reporting

The URL for this is:

  http://www.ietf.org/internet-drafts/draft-ietf-ldup-urp-07.txt

Steven reported that there are no technical changes to this
document; it was reissued to avoid becoming obsolete. Some
minor work needs to be done to update references, and then
it is ready to go.

2c. LDAP Replication Architecture - Chris Apple reporting
                                    for Uppili Srinivasan

The URL for this is:

  http://www.ietf.org/internet-drafts/draft-ietf-ldup-model-08.txt

Chris reported that there are no technical changes to this
document; it was reissued to avoid becoming obsolete. Ed Reed
is retiring as an author from this effort, but John Merrells
and Gerry Maziarski have both volunteered to help. While the
document is ready from the point-of-view of the LDUP requirements
and the Mandatory Replica documents, it still has references to
many issues that are outside the scope of LDUP as it is currently
defined. Examples of this include access control and the
administration model. Thus, the plans for finishing this
document are to first ensure consistency with other LDUP
documents, then to excise these other issues, and finally
to do one last pass before a last call.
	
2d. LDUP Replication Information Model - Chris Apple reporting
                                         on behalf of the authors

The URL for this is:

  http://www.ietf.org/internet-drafts/draft-ietf-ldup-infomod-06.txt

This is an important draft, especially because many other documents
depend  on it. An update for it is promised by mid to late April.
The current update  was done to avoid it expiring.

2e. The LDUP Replication Update Protocol - Chris Apple reporting
                                           on behalf of the authors

The URL for this is:

  http://www.ietf.org/internet-drafts/draft-ietf-ldup-protocol-04.txt

Chris reported that there are no technical changes to this document;
it was reissued to avoid becoming obsolete. It is in good shape for
being released. 

Two actions need to be taken before that can happen:

1) Document Editor needs to complete search of mail archives to
ensure that no open issues exist (currently doesn’t believe that
any exist).

2) Changes to other LDUP docs that affect this document will
of course cause it to change; we’ll need to see if any such
changes happen or not.

2f. General Usage Profile for LDAPv3 Replication - Chris Apple reporting
                                                   on behalf of the authors

The URL for this is:

  http://www.ietf.org/internet-drafts/draft-ietf-ldup-usage-profile-04.txt

Chris reported that there are no technical changes to this
document; it was  reissued to avoid becoming obsolete.
The co-chairs would like to see comments on this document.
One thing to think about as you comment is whether this
profile is sufficient, or whether the single- and multi-master
replication profiles are still needed.

2g.  Mandatory LDAP Replica Management - Chris Apple reporting
                                         on behalf of the co-authors

The URL for this is:

  http://www.ietf.org/internet-drafts/draft-ietf-ldup-mrm-02.txt

This draft is dependent on the information model, and can’t
really be updated until the Information Model is updated.
It will be turned around as quick as possible once this is done.

3. WG Charter Revision/Review

The only thing that has changed so far was the milestones. This
is because we have established consensus on a path to conclude our
work. If the co-chairs detect that there is significant slippage
in these dates, then the co-chairs will most likely recommend
that the working group conclude immediately.

We need to discuss on the mailing list whether we need to keep
the single- and multi-master profiles in the charter, or whether
the general usage profile is sufficient.

Kurt expressed concern over the charter wording, and would like
to see  "re-evaluate" changed to "conclude." Chris responded
that in the past, the use of "conclude" was discouraged by the ADs.

4. Next steps

Chris will revise the charter posting.

The LCUP gang will get together with Kurt and report progress
(or lack thereof) to the WG by this Friday (3/21).

Information model draft needs to be discussed and revised
as soon as possible.

End of meeting.