Minutes of the Third IETF-Directory Services (OSI-DS) Working
Group
February 12-13, 1991
Menlo Park (SRI)
1. Attendees
Mark Knopper Merit 313-763-6061 mak@merit.edu
Tim Howes U-Michigan 313-764-2278 tim@d.cc.umich.edu
Alex Pepple BNR 613-763-7684 alexp@bnr.ca
Peter Mierswa DEC 508-486-5581 mierswa@smaug.enet.dec.com
Bill Nowicki Legato 415-329-7856 nowicki@legato.com
Russ Wright LBL 415-486-6965 wright@lbl.gov
Arlene Getchell ESnet/NERSC 415-423-6349 getchell@nersc.gov
Ruth Lang SRI 415-859-5608 rlang@nisc.sri.com
Jose Garcia-Luna SRI 415-859-5647 garcia@sri.com
Steve Kille UCL +44-71-380-7294 s.kille@cs.ucl.ac.uk
M. T. Rose PSI 408-562-6222 mrose@psi.com
Barry Holroyd Sun 415-336-2949 berries@eng.sun.com
Paul Koski HP 408-447-3461 koski@hpindeg.cup.hp.com
Peter Yee NASA 415-604-3812 yee@ames.arc.nasa.gov
Chris Weider Merit 313-936-2090 clw@merit.edu
Stuart Cain HP 408-447-2417 scain@hpindeg.cup.hp.com
Cyrus Chow Nasa 415-604-6843 cchow@ames.arc.nasa. gov
2. Agenda, Revised
a. Agenda, Revised
b. Minutes of previous meeting
c. Liaisons: RARE WG3, NIST, NADF, AARN, ANSI
d. Replication
i. Replication Requirements
ii. Replication Solutions
iii.Network Addresses
iv. Presentation Addresses
e. APIs for the Pilot
f. User Friendly Naming
g. Domains and X.500
h. Representation of Network Info in X.500
i. DSA Naming
j. Building Internet Directory/Strategy
k. Operational Pilot Status
l. Monthly Reports on Pilots
m. New working groups: Operations, User Support
n. Internet Schema
o. Naming Guidelines
p. Naming for Internet Pilot
q. Security
r. Directory Assistance Protocol
s. Quality of Service
t. Date and Venue of next meeting
3. Introduction
The meeting was opened by Steve Kille at 9:10am on February 12,
1991. The agenda was slightly revised and massively reordered.
4. Minutes of Previous Meeting
Steve thanked Richard Colella and Peter Whittaker for producing
the minutes. He reported on the status of some of the action
items at the last meeting. The formatting of the documents has
been improved. The "Infrastructure" document met with some
difficulty in forwarding as an RFC. Steve was asked to produce a
separate "Strategy" document and to revise the RFC. Steve
contacted Al Grimstad to check on a user friendly naming related
proposal, and found that this is no longer relevant. There were
no corrections to the minutes.
5. Liaisons
a. RARE WG3
Steve reported on this meeting which took place in Brussels in
January. They discussed the activities of our IETF-DS group.
Their next meeting is April 16-17 in Utrecht, Holland. They meet
three times per year. They are very interested in getting more US
participation. Future meetings are in July, and also October 31-
November 1. Can the IAB find funding for international travel for
IETF members? Steve will look into the funding question with
appropriate people. European meetings usually have 1-2
representatives from each country. They would also like
representation from the FOX project.
b. NIST
Stuart Cain reported on the Directory SIG meeting in December.
They discussed implementation agreements for replication and
access control. They would like to see the requirements from our
group. NIST is working from the current CDAM. There is already a
stable implementors agreement based on the 1988 CCITT
recommendation. The new spec is expected by the end of the year.
The next meeting will be in March. Steve has replied informally
to the NIST liaison to encourage coordination between the two
groups and also to share our documents on replication
requirements and solution. The sense of this was agreed to by the
group, and it will be used to generate a formal liaison response.
The NIST group is concerned with "freezing" their agreements
based on a DIS version of the standard, and will be working to
avoid that kind of discrepancy.
c. North American Directory Forum
Marshall Rose reported that the last meeting was in October,
before the last IETF-DS meeting. The next meeting is in March,
after this meeting. Oh well.
d. Australian Academic Research Network
Steve received a liaison statement from George Michaelson.
Standards Australia is working on X.500 naming and addressing
standards. They will send people to the IETF some time this year.
They have not been able to participate in this group due to lack
of funds.
e. ANSI US Directory Ad Hoc Group
Roy Van Dorn (HP) reported that this group met last week. They
are bringing ballot comments to ISO. Subordinate references will
be replicated, according to the latest draft standard.
Replicating cross-references will not occur. Hoyt Kesterson is
the ISO Rapporteur. Skip Sloan will be the head of the US
delegation. Steve will send them the replication documents from
our group. There will be one more US meeting in March for ballot
comments. The liaison of the group's documents to ISO will be
done through ANSI by Paul Koski. Access control and replication
are US priorities. Some of the schema document will get into the
1992 standard. The definitions of attributes will be more like
1988. The four types of object classes will continue. Subtrees
and partial entries within subtrees can be replicated. A
completeness flag is included in replication. Searches on
attributes that don't exist will be referred for further lookup.
The unit of replication is an entry, not an attribute within an
entry.
6. Replication
a. Replication Requirements
It was agreed that this Internet Draft (Replication Requirements
to Provide an Internet Directory Using X.500) be progressed to an
RFC.
b. Replication Solutions
There was substantial discussion of this paper. Marshall and
Steve revised the text during the meeting and redistributed the
document. Marshall suggested that the title be changed to include
the changes to Distributed Operations as well as replication.
This suggestion was agreed to by all. A number of changes were
suggested to make the document more clear. There was a suggestion
to include a figure describing knowledge replication. None of the
proposed changes require discussion at a further meeting, and
Steve agreed to send a revised document out to the list on Monday
(February 18). The group will respond within one week with any
comments. After that the Internet Draft (Replication to Provide
an Internet Directory Using X.500: A Proposed Solution. However
the title may be changed.) will be progressed to an RFC.
c. Network Addresses
There were a few comments from the IAB regarding the Telex
kludge. It was agreed that this Internet Draft (An Interim
Approach to Use of Network Addresses) be progressed to an RFC.
d. Presentation Addresses
It was agreed that this Internet Draft (A String Encoding of
Presentation Addresses) be progressed to an RFC.
7. APIs for the Pilot
Ruth Lang said that this was an important area and would like to
see suggestions for APIs (application programming interface). The
only comment received so far on the list was from Peter Whittaker
(BNR) about object management support in XOPEN. There was a
discussion of the XDS agreements. Peter Mierswa said that DEC
participated in XDS. The user-friendly and object-oriented
aspects of XDS will cause applications to be large. It is
difficult to extend the XDS object set. There are other technical
drawbacks, but it was agreed to by a number of parties. DEC will
support the XDS API but also a more functional layer. Quipu does
not support XDS. XDS and object management documentation is
available from Omnicom. It was felt that APIs did not fit into
our group's charter. We may want to make recommendations but then
move on to the technical infrastructure. This group is also not
to manage projects or pilots.
8. User Friendly Naming
Peter Mierswa tried to find a common syntax set with the OSF DCE
naming (based on unix filesystem syntax) and the proposed X.400
annex for business card OR address format (uses semicolons and
slashes, which evolved out of the RFC 987 work). However there
was no such syntax in common and Peter gave up. The algorithm in
this document is useful based on experience, though there may be
scope for experimentation. It was noted that name space
organization affects efficiency of searches. For example
Cambridge University uses many levels of OU. It is recommended in
the Naming Guidelines document (see section 18) that pilots be
laid out so that this user friendly naming scheme works
reasonably. It was agreed that this Internet Draft (Using the OSI
Directory to Achieve User Friendly Naming) be progressed to an
RFC.
9. Domains and X.500
UCL has done some work in implementing this scheme. There is a
tool to do a white pages lookup based on a domain address. This
is an experimental service. The general appropriateness of
representing domain name system information in the Directory was
discussed. This is viewed as controversial. The X.500 version of
DNS may have be usable for other functions than those currently
offered by the DNS, such as browsing. Mailbox records are
included in the DNS, but are not widely used. Peter Mierswa said
that it would not matter if this was not submitted as an RFC.
Steve disagreed with that and would like to progress the work.
Tim Howes suggested that we submit this with a disclaimer that it
is experimental. Steve would like the IAB to discuss these
issues. Jose Garcia-Luna felt that security should be discussed
in this paper. It was eventually agreed that this Internet Draft
(Domains and X.500) should be progressed as an RFC.
10. Representation of Network Information in X.500
Mark Knopper and Chris Weider gave a presentation on some work in
progress at Merit, which will become part of the DARPA/NSF
sponsored Field Operational X.500 (FOX) project. They have
entered the network contacts part of the whois data into the
@o=Internet part of the White Pages DIT. New object classes have
been defined. Bill Nowicki noted that putting all of the IP
network numbers into a single location in the DIT will not scale
well. It was suggested that the network number entries be located
within the owning organizations. This would obviously require
much more participation in the X.500 projects. For now the net
numbers can be entered in a separate tree under o=Internet and
eventually these entries will just be pointers to the master
network entries. Steve proposes another solution to this in the
Domains and X.500 paper. It is scalable, but also requires more
work to bootstrap. There will be further cooperation with SRI,
ISI and PSI to allow the rest of the NIC's data to be entered
into X.500. There were a number of useful suggestions on how the
network information could be stored in the DIT. It was
recommended that Merit produce an internet draft to document this
effort, both work in progress as well as long term design. Chris
agreed to do this by March 7. He will take the scalability issues
into account.
11. DSA Naming
The current South American wildlife names don't seem to be
descriptive enough! The solutions outlined in this paper solve
some operational problems with quipu-based pilots. Peter pointed
out that the section on multinational organizations does not
solve the problem. There were several suggestions for
modifications, and discussion of this will be necessary at the
next working group meeting. It was felt that after that, this
Internet Draft (DSA Naming) can be progressed to an RFC.
12. Building Internet Directory/Strategy
The infrastructure Internet Draft was held up in protracted
discussion regarding how to submit RFCs. Steve wrote a new
strategy document. It was agreed that APIs should be mentioned in
this document. The "strategy" was removed from the I.D. and so
that was renamed to a very long name beginning with "Overall
Plan". The strategy document was agreed to in principle but will
not be forwarded at this time. The Overall Plan Internet Draft
was agreed to be progressed to an RFC again.
13. Operational Pilot Status
a. PSI Pilot
Marshall reported that there are about 70 organizations on the US
pilot. Growth has been linear since the pilot began. ISODE 6.8
interim release is due out by the end of the month. It is a very
stable and higher performance version. It will have Tim Howes'
mods to quipu, and also the Directory Assistance Protocol (which
allows splitting the DUA between two different hosts). FRED is
faster now. There is a Macintosh DUA offered by PSI as shareware.
A source license is available similar to the Nysernet SNMP
license. The PSI pilot only allows DSAs to be connected via IP
(and now CLNP). The quality of X.25 in the US "sucks dead pigs
through a straw". [Ed. Note: It has been suggested offline to
formalize this language to "provides pneumatic inward pork-
pressure via narrow flexible tubing".]
b. COSINE Pilots
Steve reported that 19 out of 20 countries in COSINE are running
X.500 pilots. The COSINE P2.1 pilot has been renamed as PARADISE,
and has officially started. Its manager is David Goodman. ULCC
has an operational facility to replace Giant Tortoise. Their plan
is to support international pilots until the end of 1992. France
has a research pilot based on quipu and also a commercial pilot
based on Pizarro. Xtel and the Dutch PTT are involved in
PARADISE.
14. Monthly Reports on Pilots
It is felt that the operational pilots should distribute status
reports on a monthly basis. The FOX project is interested in
coordinating the US report. Ruth Lang contacted Jon Postel at ISI
about this and Jon volunteered ISI to produce the reports. Some
FOX mailing lists will be set up to help coordinate the US
report. David Goodman, the PARADISE manager, will integrate this
into the international report. FOX and PARADISE will agree on
timescales for ensuring that this comes out each month. Reports
will be timely, with noncontributors marked as "no report for
XXX". This international report will be sent out as a part of the
Internet Monthly Report and to a separate list for those not
interested in other aspects of the IMR. The reports should be on
"The State of the DIT". Organizations should be queried for their
activities. Marshall gets regular statistics reports from the US
DSAs. The Canadian pilot is operated by the University of
Toronto.
15. New Working Groups
a. X.500 User Support Working Group
Chris Weider volunteered to chair a new working group. Steve will
talk to the IETF area coordinators and suggest that the new group
be jointly in the OSI and User Services areas. Several of the
group participants were interested in joining the new group. The
first meeting will be at the next IETF. Chris distributed a draft
charter and several comments were made. Chris will talk to Joyce
Reynolds and Dana Sitzler, to see whether it would be reasonable
to model the group after the NISI working group. Perhaps the new
group should be called DISI (pronounced "dizzy"). The group would
provide a documentation package for sites, as well as a center of
expertise for X.500 issues.
b. X.500 Operations Working Group
There was some interest in forming such a group but it was felt
that this should wait until the activities of the main IETF-DS
group come to an end, or at least go into "maintenance mode". It
was viewed that the group will only last for one more meeting
with the same high level of activity. After that the operations
group will be formed. Marshall Rose and Chris Weider were
involved in discussing the charter of the new group.
16. Internet Schema
Marshall suggested that the name of the Internet Draft (COSINE
and Internet Naming Architecture) be changed from "naming
architecture" to "schema". This was accepted. There were comments
on this document at the RARE WG3 meeting. The
textEncodedORAddress attribute was deprecated by OSI purists, but
some members felt it was useful in the pilots. This Internet
Draft was agreed to be progressed to an RFC.
17. Naming Guidelines
Steve introduced this Internet Draft and explained that it sets
out some guidelines for how to lay out a pilot DIT. It is a
followon to annex B of X.521. Marshall mentioned that the T.61
character sets for international symbols once were a problem but
work now in quipu. Peter mentioned that this is not a solution
for multinational organizations. It is viewed that this is a
difficult problem, and that the acceptable solutions should be
documented. There needs to be a definition of "multinational
organization". HP would like to see a single "mount point". There
was a discussion of organization naming strategy. Marshall
suggested that the names be fully descriptive to avoid later,
possibly legal, conflicts. The naming authorities must enforce
unique names within the DMD. Long names were recommended.
Marshall mentioned that a small DIT depth makes browsing less
effective. It is not useful to define conformance rules for a
guidelines document. Conformance is useful for a given national
pilot. Steve and Paul Barker will edit the document and
distribute to the group. At the next meeting it will be proposed
that the Internet Draft (Naming Guidelines for Directory Pilots)
be progressed to an RFC.
18. Naming for Internet Pilot
Marshall gave a presentation of a paper he and Einar Stefferud
had written to be presented at the NADF, US-CCITT-Study Group D,
and ANSI as well as to this group. The problem is that there are
no OSI numbering authorities in the US, but they are needed for
pilots to advance to a production stage. ANSI has accepted over
500 applications for OIDs under 1.2.840, but due to legal
problems have not assigned any. Numbers are not a problem for
ANSI but names are. The only legal method would be to assign the
name and then publish the fact in the Federal Register with the
reserve to revoke on a 6-month challenge procedure basis. GSA has
been assigning NSAPs under AFI/IDI=47/0005, only for federal
agencies. IANA has assigned several hundred OIDs under
1.3.6.1.4.1 for internet network management use. US-CCITT-SG-D is
trying to make a national decision on naming, but only for an
X.400 ADMD/PRMD registry and not for X.500. Possible naming
universes are geographical, political or community. Civil
authorities are the best choice as it gives a familiar and
undisputed structure. However collisions in RDNs must be avoided.
The proposal suggests using the numeric code assigned by ANSI for
the RDN itself. This was heavily disputed, but as Marshall noted
it would be legally defensible. The consensus was that we should
fix ANSI rather than using numeric RDNs. Marshall and Stef
believe that their presenting this proposal to the four groups
will force a national decision. The proposal went on to recommend
use of numeric codes for states and populated places. Naming of
OSI entities was included, and there was a suggestion that non-
OSI entities should get names too (eg. SNA, TCP/IP applications).
Steve suggested that this be made into an Internet Draft but not
a standard. Marshall will make the changes suggested by the group
before the NADF presentation in March. He will "lean heavily" on
ANSI to begin assigning names. Beth Summerville is ANSI's
registrar for the naming authority function.
19. Security
Peter Yee's paper was revised since the last meeting. There were
not many changes due to lack of comments at Boulder. Marshall
said that it will be necessary to consult with the IETF Security
working group before progressing this document. Peter will
contact Steve Crocker to get help on proper security terms and
concepts. Marshall suggested splitting the discussion in the
paper between authentication (simple now, strong later), and
authorization (access control lists). Paul suggested including an
ACL to control access for searching. Steve suggested that this
should become an Internet Draft with title Security Requirements
for X.500 in the Internet. There should be a companion document
for Security Solutions, and this should reference the 1992 CCITT
document. A problem at MIT is that they want to limit searching
their organizations to return data only if less than n entries.
HP wants to disallow searching their organization entirely. Peter
will revise the document and send it out to the list by March 1.
20. Directory Assistance Protocol
Marshall wrote an RFC describing a protocol used by PSI's
Macintosh DUA client. It documents existing practice and is not a
standard. The server is part of ISODE. He characterized the
protocol as "horrid". Tim Howes has also been working on a
Macintosh DUA with a different protocol. Tim will write an RFC
for his DAP pretty soon.
21. Quality of Service
Steve submitted an informal writeup to suggest that QOS
attributes be added to the schema to represent the advertised
quality of DSA services in the pilots. This was thought to be a
good idea and there were no objections to including this in the
Schema document.
22. Notable Actions, Dispositions and Promises
a. RFC Progression
The following documents were recommended to be progressed to RFC
status:
Replication Requirements to Provide an Internet Directory Using
X.500 (section 6a)
Replication Solution and Distributed Operations (section 6b)
An Interim Approach to Use of Network Addresses (section 6c)
A String Encoding of Presentation Addresses (section 6d)
Using the OSI Directory to Achieve User-Friendly Naming (section
8)
Domains and X.500 (section 9)
Overall Plan (section 12)
Internet Schema (section 16, and including QOS item in section
21)
Naming Guidelines for Directory Pilots (section 17)
b. Action Items
Strategy document will be revised by Steve (sections 4, 12). The
issue of travel funding will be investigated by Steve (section
5a). A formal response to NIST will be drafted by Steve (section
5b). The replication documents will be sent to ISO via ANSI and
Paul Koski by Steve (section 5c). Jon Postel, for the FOX
project, will set up a mailing list, and produce monthly reports
coordinated with PARADISE and the Internet Monthly Reports
(sections 10 and 14). Chris Weider will start up the new
Directory Information Services Infrastructure working group
(section 15a). Chris and Mark will write an RFC on representing
network infrastructure information by March 7 (section 10).
Marshall Rose will lean heavily on ANSI to assign organization
ids and names (section 18). The security document will be revised
by March 1 by Peter Yee (section 19).
23. Date and Venue of Next Meeting
There will be no OSI-DS meeting at the March IETF. The next
meeting will be after that, to be decided on the list. A
possibility is a video conference, or alternatively a face to
face meeting either in Ann Arbor or on the east coast in May or
June. The choice depends on online discussion of the working
drafts. Given some comments, it might be appropriate to wait
until July. Steve will poll the group after the next round of
editing.
23. Thanking the Host
Ruth Lang and SRI International were thanked for their excellent
services including a lunch.