DirDep BOF, 39th IETF Munich
Wednesday 9-11:30
Written by Sally Hambridge
All typos and mis-understandings mine alone.
SUMMARY:
The DIRDEP BOF met and decided to concentrate on LDAP Service Deployment
as a way to focus activities. Accordingly, the proposed Working Group
will be changed to Ldap Service Deployment, with the acronym LSD.
There was a large scoping disccussion where we decided on the
size of the problem we were willing to tackle, and then we prioritized
the items of work. Finally we assigned document editors for the
documents.
Minutes:
Chairs: Chris Apple, capple@master.control.att.com
Roland Hedberg, Roland.Hedberg@umdac.umu.se
Mailing List: ietf-dirdep@listserv.umu.se
Subscribe: listserv@listserv.umu.se
In body: subscribe ietf-dirdep <full name>
Archive: ftp://ftp.umu.se/ietf/dirdep/archive
As ASID and IDS are winding down, there is still a need for work in the
Directory area to define differences in DITs and schema, to handle
I18N (Internationalization) problems, and deal with character set
problems. Also, approximate matching algorithms such as soundex
and metaphone do not work well for non-English names; there are
schema inconsistencies in attribute names and semantic
collisions across schema exist in abundance. The general problem
of locating a directory service has not been documented; IWPS
is not really a true schema if a schema is defined as attribute name,
semantics, syntax, and inter-relationships. Guidelines for schema
writers do not exist as an IETF document, directories can be difficult
to administer, maintain and operate, and from the user's perspective,
the same query as entered in different clients, received different
results from the same server since "guessing game" search filters
are contained in clients.
The problem we are trying to solve it: How do I find a service to find
information about people; and How do I find information about people.
The group elected to concentrate on LDAP implementations to limit the
scope of what the working group was biting off, and therefore changed
the name of the proposed working group to : LDAP Service Deployment,
which gives the interesting acronym of LSD. (A natural follow-on from
ASID). Accordingly, the mailing lists sited above are slated to change
the week after the 39th IETF to:
Mailing List: ietf-lsd@listserv.umu.se
Subscribe: listserv@listserv.umu.se
In body: subscribe ietf-lsd <full name>
Archive: ftp://ftp.umu.se/ietf/lsd/archive
We agreed we would be LDAP focused but not afraid of dealing with other
protocols which help directory deployment such as DNS, and CIP.
There was a fairly long discussion on scoping. The original suggestions
were - schema inconsistencies and mappings; piloting activities in support
of particular application use of a directory service (e.g. forward knowledge
of certificates and global white pages service); help and guidance for the
people who deploy directories; help and guidance for people building
implementations. Other suggestions from the chair: delpoyment and
requirements for mission cirtical directories, general problems of
locating directory services.
We had a long discussion on an area which was originally described as
"managing the root context" or "interconnecting ldap islands" but was
finally named "practical guidance for naming and interconnection.
We elected to call this out as a separate item of work from "giving
guidelines to implementors" so it would not be lost or forgotten
in the shuffle.
After discussion and a show of hands, the scope and priorities of the
proposed Working Group will be:
H = High Priority
M = Medium Priority
L = Low Priority
H - practical guidance for naming and interconnectivity
H- - document schema inconsistencies and mappings
M+ - provide help and guidance for the people who ddeploy directories
M - document guidelines for locating a directory service
M- guidelines for implementors and developers
L - support pilot for particular application use of directory service
There was some discussion of where character set issues fit, would it
be in this group or in the LDAPExt (LDAP Extensions) group, and Chris
said he and Roland would talk to Mark Wahl and Tim Howes about this.
There was a discussion about whether the nature of the pilot would
be and Internet or Intranet pilot, and the group agreed it would be
an Internet pilot.
We then decided on the documents to support the work and document
editors for each:
Charset Handling - to be negotiated with LDAPExt
Schema inconsistencies and mappings - Bill Curtin, Chris Apple
Guidelines for Schema Writers - Mark Wahl and John Strassen
Guidelines for LDAP Client and Servers Implementors - Ed Reed, Sandy Miklos
Guidelines for Naming and Interconnection of LDAP directories -
Ann Brown, Chris Weider
Requirements for successful LDAP deployment - John Strassner, Ed Reed
Requirements for LDAPv3 Pilot Study - answer sought - Probably will
be dropped from the charter
LDAP Server Mesh Pilot Plan - Left Open
LDAP deployment guide - Chris Apple, Sandy Miklos
Approximate matching algorithm - Roland Hedberg
TISDAG Project Report - Patrik Faltstrom
(TISDAG is the Swedish Directory project, and the report is at
http://tisdag.sunet.se/ )
Guidelines for locating a directory service - This work has been started
by Ryan Moats, Paul Leach and Bruce Greenblatt and the group thought
it polite to ask these 3 people before committing them to the work.
Action Items:
Chris will poll the Mailing list about the pilot study and see if
it should retain the low priority or stay in the charter at all.
Chris will post goals and priorities to the list by Monday after IETF.
Roland will change the mailing list name and keep all currrent subscribers
by the week after IETF.
Chris will send the question about LDAPv2 to v3 compatibility to the
list by Monday.
Chris will also talk to the LDAPExt charis before next week about the
character set stuff and about the relationships for locating directory
services.
A new version fo the charter will be posted 2-3 weeks from now and the
group should close on it by October.
Documents Editors should send mail to Chris and to Roland to
confirm their committment.