Subject: Minutes 43
Date: Mon, 4 Jan 1999 10:40:32 -0500
From: igorf@holta.ho.lucent.com (Igor Faynberg)
To: igorf@holta.ho.lucent.com (Igor Faynberg)

Minutes of the PINT Meeting at the 43rd IETF Conference

(Reported by Steve Bellovin and Igor Faynberg, based on the notes taken by
Lawrence Conroy, Hui-Lan Lu, and Lev Slutsman.)

The PINT WG met from 15:45 to 18:00 on December 8, 1998, with a 15 min. break
from 16:45 to 17:00.  The attendance list has registered 177 participants.
The meeting was chaired by Steve Bellovin (AT&T Labs) and
Igor Faynberg (Bell Labs, Lucent Technologies). The meeting notes were taken by
Lawrence Conroy (Roke Manor Research/Siemens, Hui-Lan Lu (Bell Labs) and
Lev Slutsman (AT&T), to whom many thanks.

Igor presented the agenda, which was accepted with no changes.

Scott Petrack (Vocaltec) made a short presentation on the two remaining PINT
issues: 1) Security section needs a volunteer to provide any additional (to that
of SIP security) material, if needed; and 2) registration of the numbers for
services. These issues are to be addressed in on-line discussions in the course
of the next two months before the final version of the draft is issued.

Lawrence Conroy (Roke Manor Research, Siemens) addressed the mapping of PINT
parameters to those of the InitiateCallAttempt operation of the Intelligent
Network Application Part Protocol (INAP), and demonstrated that those are
understandable to IN service logic. In the course of the presentation an issue
of using the ACK message bodies for carrying additional information was raised.
The discussion  that followed resulted in the decision not to use the body of
ACK in PINT. Another issue, that of the treatment of  not understandable
headers,  had a simple resolution, too: ignore such headers.

Lev Slutsman (AT&T) presented a need for notification mechanism in order to
avoid unnecessary PSTN circuit occupation when calls are queued.  If the PINT
protocol--in its present form--requests a connection  between a calling party
and, say, a call center in which all agents are busy, the calling party will be
connected (via IN) to the announcement while waiting for an agent. (The calls
that cannot be answered are normally put into a FIFO queue by IN.) There is
nothing wrong with that because this is how the Freephone service works;
however, the advantage of the Internet access can be used so as to inform the
calling party about the queue progress and then establish the call only when
there is an agent to answer it. Doing so will significantly decrease the use of
circuitry (a resource that is becoming scarce) while improving the customer
satisfaction.  With the present IN facilities, the information about the queue
size and anticipated waiting time is available at the SCF (and, subsequently,
the PINT Executive System and PINT Server). Lev's proposal was to augment the
PINT protocol with the feature by which, the PINT Client can request
notification about the queue progress from the PINT Server. He suggested that
the notification mechanism described in the recent mmusic WG draft "SIP for
Presence"  be used. There was a unanimous agreement that this feature was
important.  As far as the particular implementation is concerned,
Jonathan Rosenberg, the co-author of both SIP and the draft in question,
recommended that only the facilities of the latest SIP specification be used.
He explained that the notification mechanism mentioned by Lev was put together
for the Presence Information Protocol proposal, and was not intended to be
part of the "core" SIP. To summarize, Lev's proposal can be implemented with
the "core" SIP and the SUBSCRIBE/NOTIFY extensions that have been defined for
PINT.  Scott Petrack demonstrated that the present SUBSCRIBE/NOTIFY PINT
mechanism, augmented by filtering, is sufficient to support the feature.
There was no disagreement with that solution.

Scott expressed a concern over the way in which the PINT Client
could address the Executive System; presently, it is impossible to
to identify it. Lev agreed that the identification can be implicit (i.e. the
PINT Client only "sees" the Gateway, and that sorts out any "back end"
addressing needed).

Scott and Lawrence confirmed that the final copy of the PINT Protocol draft
(including the example demonstrating an implementation of Lev's proposal) will
be available in about two months.

That concluded the first part of the meeting agenda; the second part was
dedicated to two informational presentations and the discussion of the future
work.

Gilles Lecorgne, France Telecom-CNET/TINA-C, has presented the TINA-C status as
well as its architecture, the PINT model in its context, and the areas in
which TINA is planning to contribute to PINT. The chairs thanked Gilles and
invited further communication from TINA-C on the result of the PINT-related
prototyping work.

Alec Brusilovsky (Lucent Technologies) reported on an implementation of the
Internet Call Waiting (ICW). The presentation generated much interest. It was
noted that this implementation was particular to the needs of Local Exchange
Carriers (LECs).

Finally, Steve Bellovin addressed the future work in PINT.  He noted that the
existence of the PINT MIB is not essential to the advancement of the PINT
protocol draft to the  'proposed standard' status, which is the next milestone
for PINT, but it will be necessary for the advancement to the 'draft standard'
status.  The MIB specification work should start, however, as soon as
possible. (Igor mentioned that Dan Romascanu, the HUB MIB WG chair and a
"MIB Doctor" has volunteered  to work on the PINT MIB.) Meanwhile, the
implementations are to be developed and
reported to PINT. A request for show of hands of those who work on PINT
implementations resulted in a count of five independent implementations in
progress. While the implementations are progressing, the PINT WG may remain
dormant (for about two meetings); meanwhile the proposals for the new work
(if any) will be considered.