Met in Manhattan Room, 1300, Friday 26 March, 2010
Scott Bradner chaired.

NOTE WELL presented.
Minutes by Tom Taylor

Agenda and Status
=================

Ken Carlberg raised the issue of the fate of RSVP given the mini-BoF on its future earlier this week. Scott Bradner explained the outcome of the meeting, which is that it is likely an RSVP directorate will be formed to act as a target for work on the protocol. Scott suggested that the protocol in its present form was designed a long time ago and may need change to serve the purposes people have in mind now. Lars (as AD) noted that there was a need to have RSVP-related IDs reviewed when they get submitted to the tsvwg.  Ken asked that people volunteer to review material related to RSVP other than their own.  A number of people volunteered to do such reviews. 

No further comments on the agenda.

Scott provided an update on the status of the WG.  He noted that the WG now had 5 RFCs published and an ID in IESG evaluation. He noted that we lag far behind milestones on remaining work. No comments on status.

Signalling Requirements
=======================
Georgios Karagiannis, charts

Flow IDs: Bob Briscoe remarked that it will be important to specify the format of the list so that the other end can understand it.

Discussion of reliability requirement for flow IDs. Philip Eardley: not disastrous if it goes missing. Scott suggests: text should say that it must be received (rather than sent) reliably. Tom Taylor noted that there is a hierachy of reliability requirements: notification of excess-traffic-marking is most important, flow IDs less important, routine reports, no excess-traffic-marking are least critical.

Scott asked if draft had any requirements on authentication and the like. Response - requirements are present in the slides, but not yet in the draft.

Requirements for signalling between egress nodes and decision point: 
--------------------------------------------------------------------

Last bullet of first slide: wanted clarification of requirement for node identifier. Identifies node that sent the information. Issue of visibility of information: why should it not be visible outside domain, need to rephrase because technically PCN domain would not include a centralized decision point.

On reliability: state requirement, but not the mechanism. Some discussion of the wording of the "PCN reporting frequency" bullet. State that for regularly reported information, don't need an additional reliability mechanism. For other information you do. 

Security considerations -- may need protection against replay attacks. Under confidentiality: operator may not wish messages to be visible outside the control domain. Scott observed that integrity is probably not so important as authentication, and in any event integrity comes for free with the latter.

Filter spec: Bob Briscoe suggests referring to another draft, Scott suggests keeping it more generic. 

Requirements for signalling between ingress nodes and decision point: 
--------------------------------------------------------------------

Same remarks apply.

Support to be a WG document amongst "those who care". Will put call out to mailing list to confirm. Publish revision in the meantime. Philip Eardley: tidy up structure. Scott: put discussion of reliability in a single place.

HOSE draft
==========

Georgios Karagiannis, charts

To the list for determination of further steps.

CL and SM Edge Behaviour Drafts
===============================

Tom Taylor, charts

Make activation of flow admission also configurable.

Make ratio between Tmax and Tfail 1:3 as Philip Eardley suggested on the list.

No opposition to WGLC one new version is issued.

Piggybacking Draft
==================

Tom Taylor, charts

Agreed that the proper course of action is for Tom Taylor to work with Francois LeFaucheur to complete the 2006 draft. The problem of the appropriate target for RSVP messages relating to flow termination was seen as valid, but recent Baker and LeFaucheur drafts may have provided the necessary extensions to deal with collections of flows. Ken Carlberg may have references.

Tunneling and ECN
=================

Bob Briscoe, charts

Would like support on the TSVWG list.

Encoding Comparisons
====================

Bob Briscoe, charts

Read -03 version to allow advancement to WGLC.

Philip had comments that should be taken into account in -03. In effect, doesn't capture flavor of why we made the decisions we did.

Other Business
==============

Bob Briscoe: if tunneling draft goes through, the PSDM encoding draft may be moot. The other two may be ready for WGLC.

Clarifications will be added based on comments received.