Draft GSMP minutes, IETF 56

General Switch Management Protocol (gsmp)

Thursday, March 20 2003 at 1300-1500
=====================================

CHAIRS: Avri Doria <[email protected]>
        Kenneth Sundell <[email protected]>


Minutes: Fiffi Hellstrand <[email protected]>


1. Agenda bash

   Nothing changed.


2. WG Status:

   Avri went through the status of the GSMP working group and the 
   SubIP area. The IESG decided to not ask NomCom for a permanent 
   chair of the SubIP area. SubIP area will eventually be closed 
   down and those working groups that not soon will end will go to 
   other areas. A proposal is that GSMP goes to Operations and 
   Management if not ended before closing SubIP. Scott Braden has 
   resigned as AD for SubIP and Bert Wijnen will take over the 
   responsibility of the GSMP WG.

   Comments/Questions from the floor:
   Comment 1: Good idea with Operations & Management Area. 
   Comment 2: Why not Traffic Engineering? 
   Avri: Leave it to TE WG to decide.

   Avri continued with the GSMP WG status. The GSMP WG is behind its 
   schedule and several of the milestones are outdated. Expect to go to 
   Proposed Standard this year with all work but it will take another 
   6 months before getting to Draft Standards due to development and 
   implementation progress. GSMP WG may very well go dormant between 
   these activities. 
   There is no proposal or activity to align with Forces WG. 
   It will be up to the Forces WG to take on any of the GSMP work. 
   According to their schedule, the Forces documents are still open 
   until Vienna.

   Q: Which protocols will or may be proposed for Forces? 
   A: GSMP is a good protocol candidate. It will not be proposed from 
      the GSMP WG as an alternative. That means that GSMP WG will not 
      take responsibility of describing how the protocol meets the 
      Forces requirements etc. It does not out rule the Forces WG to 
      choose GSMP. 

3. Requirement docs:

   -draft-ietf-gsmp-dyn-part-req-03.txt
   -draft-ietf-gsmp-reqs-05.txt 

   Avri reported that the Dynamic Partitions Requirements, 
   draft-ietf-gsmp-dyn-part-req-03.txt, has been approved by IESG. 
   It was resubmitted twice and it is now sent to the RFC editors 
   queue as informational. 

   Ken gave a report on the Optical Requirements specification, 
   draft-ietf-gsmp-reqs-05.txt. A new version was posted after 
   ID-NITS review. There are still some small editorial nits. 
   A new revision has to be submitted by the design team. 
   The new revision will then be passed on to the IESG.

   Q: Why are the requirements specifications informational? 
   A: Only protocol and features can be other than informational.

4. Specifications:

   Base Spec (Avri Doria)
   -draft-ietf-gsmp-v3-base-spec-01.txt

   Avri went through the few changes made since the initial version. 
   Results & code field has been updated. The Result field has been 
   decreased to 4 bits and the Code field increased size to 12 bits 
   in order to accommodate more Codes. The number of possible results 
   was predicted to fit into 4 bits. A new definition of Label Type 
   and Service Type ranges has been introduced. The specification 
   also includes a generalized discussion for port that support virtual 
   path. The first idea was to move the original text from base document 
   to specific ones but instead more generalized wordings were used. 
   The same applies for the connection merge functionality. The section 
   on traffic control flags has been restored after it was accidentally 
   removed (section 10.4).

   Still to be done:
   Continue removing media specific references; e.g. ATM on page 52.
   Make changes to support general requirements in <draft-ietf-gsmp-reqs-05>. 
   Continue working with optical and TDM design teams to make sure base 
   specification contains support they need. Support dynamic partioning 
   need to be put in place.

   Jonathan Sadler: Last meeting ideas of using layering. Will work on 
   putting in a proposal on that a.s.a.p.


   Packet Spec (Ken Sundell)
   -draft-ietf-gsmp-packet-spec-00.txt

   There are still some holes in the draft that needs to be fixed. The 
   document support both L2 capable and packet capable switches. The 
   intention is to add as few changes as possible into the spec to not 
   affect existing implementations.

   Main issues: restructure failure response according to the new base 
   specification version.

   Got working group consensus at the Atlanta meeting to bundle together 
   L2 and Packet capable switch types. Still need to check consensus on 
   the mailing list. 

   Messages and Types belonging to this draft:
   -Labels types and Label Range messages for ATM, Frame Relay, 
    MPLS Generic and FEC Labels

   -Add ATM specific procedures related to Connection Management 
    (for Add Branch, Move Output Branch and Move Input Branch messages)

   -In Port Configuration messages, define Port Type Specific data for 
    ATM, Frame Relay, MPLS Generic and FEC's

   Are there any opinions on the bundling of L2 and packet media? 
   The document will not be too long and a split not needed. Further 
   discussions on the list, i.e. need consensus. 

   What's next? Need to coordinate with authors of base spec to remove 
   redundant content. A new version will be submitted shortly after the 
   San Francisco meeting.


   Optical Design Team (Jun Kyun Choi)
   -draft-ietf-gsmp-optical-spec-01.txt

   Jun Kyun Choi gave a brief progress report. A new version was submitted 
   prior this meeting. In order to allow support for GMPLS, it is needed 
   to understand the encoding rules. 

   Q: What is the relations between optical and TDM? Is the overlaps with 
      regards to encapsulations? 
   A: No. 

   How is the progress of the draft? Jonathan S will send out comments on 
   the list. 


   TDM Design Team (Jonathan Sadler):
   - No I-D published yet.

   Jonathan got a question about the plans for a TDM draft by the chairs. 
   Jonathan told that the draft is not written yet but the intention is to 
   submit text to base specification first in the next few weeks and then 
   work on the draft. 

   Going forward with the specifications? Plans are to submit all four 
   specifications to the IESG as a bundle but it is not a requirement 
   if one of them will drag in time.


5. Milestones

   The main question is when these four specifications will be done. 
   Go to Vienna with four solid drafts with a list of what's still to 
   be done (open issues).

   The list of milestones has to be updated. A number of sub-steps were 
   proposed to get better granularity in the planning.
   - Layer adaptation added to base spec. Target: End of April
   - Dynamic partitioning + general change. Target: End of April
   - First version of TDM spec.  Target: Mid May
   - Packet specification update. Target: End of April
   - Update of Optical specification. End of April
   - WG last call. Target: Jun 03
   - IESG submission. Target: Aug 03

   A 6-9 month gap would be needed to get time for implementation and 
   deployment. That gives an update of submission of GSMPv3 in June 04.

   Comment: Dynamic partitioning will take some time so afraid the schedule 
   is too aggressive. The IESG submission will be earliest in August. 
   Avri: 2-3 months gap needed? Also the MIBs would need updates to support 
   the protocol. The GSMP WG needs a volunteer. 

   Comment: Good to have optical and TDM to be aligned to get the same 
   structure.

Meeting adjourned.