Sixty-six IETF Montreal July 2006

Monday, July 10, 2006
0900-1130 Morning Session I
RTG     ccamp     Common Control and Measurement Plane WG

CCAMP Working Group Meeting Report
Note takers: Emmanuel Dotaro, Lyndon Ong, Stephen Shew, Martin Vigoureux
Chairs: Adrian Farrel, Deborah Brungard

=========================================================
0. Administrivia (chairs)
Slides

One change in the agenda, JP will not present on CE-CE TE.
=========================================================

================================================
1. WG status, RFCs, drafts, milestones (chairs)
Slides

Adrian: No RFC published since Dallas, but pay attention to RFC4461 on P2MP requirements, covers both packet and non-packet P2MP. Within the next month, two bis with minor changes will be published.
Dimitri: Reminded RFC4558 on Node ID based Hello was published as Informational.
Adrian: GELS status: No further BoF. Deborah and I discussed with the ADs. People are interested to do this, and CCAMP already includes GMPLS control of Ethernet. Ethernet data plane specifications do not belong to IETF. We will look at signaling and routing protocols extensions only when there is a clear specification of the Ethernet data plane to be controlled.
We may want to do a charter modification to make things really clear.
Ross Callon: Would be ideal if agreement on a single data plane profile. Today, it seems unlikely, but we can live with a small number.
Adrian: What is small?
Ross: Ten is not, two is.
Kireeti: Not clear why we are working on GELS. What is the motivation for GELS? For the data plane, it may be two or three profiles, but each leading to different signaling profiles. So not only a question of data plane profiles but also different label formats (e.g. 12/24 bits is different from 60).
Dimitri: GELS did describe operational advantages and requirements. Difficulty to have (data plane) profile from other SDOs, is a partial reference to a document sufficient? Does other SDO need to validate? Is liaison needed?
Adrian: That would be enough to start, maybe not to finish.
Adrian: We will ask the GELS list if they want to continue with the GELS list.
Kireeti: You should also ask CCAMP.
Adrian: It will be asked on the lists.
================================================

==================================
2. ITU-T and OIF progress report (Lyndon)
Slides

Richard: Is there a liaison between ITU and IETF CCAMP on G.7716?
Adrian: There were no discussions at ITU about doing a liaison.
Loa: We will have more detailed reports on T-MPLS in MPLS WG and PWE3.
Adrian: What is required from CCAMP on the OIF liaisons?
Lyndon: They are discussion on functional requirements not protocol specification. Documents sent for information mainly, but comments welcome, especially on inter-working.
Deborah: When is the next OIF?
Lyndon: July 31st
Gert Grammel: What is the scope of E-NNI routing? Isn't it a collection of what has already been done by ITU?
Lyndon: It is requirements plus joint ITU/IETF work on ASON routing protocol extensions plus
prototyping documentation.
Gert: Then what is really the scope? Is it a recommendation?
Lyndon: Mainly documentation. Features and extensions are from ITU-T/IETF.
Zafar: How does interworking document relate to interworking draft of CCAMP?
Deborah: CCAMP's interworking is MPLS/GMPLS, the OIF is on GMPLS/OIF.
Lyndon: Actually G.7713.2/RFC3473. It's not the same as the addressing draft which is GMPLS related.
Dimitri: (referring to slide on Ethernet) What is the rationale for "large numbers"?
Stephen: Relates to the bundling feature in MEF and G.8011.x services definition, maps VLAN customer space to the VLAN in the provider space.
Dimitri: Please send requirements to the list.
Adrian: I agree.
==================================

=======================================================
3. Addressing Draft (Richard)
   - Update on status
   - Any remaining issues?
   - Reached stability?
Slides

Adrian: There has been lots of discussion whether this should be Standards Track, Informational, or BCP.
I believe this is Standards Track as it enhances the RFC.
Lou Berger: Does it gives directions, clarification, proposes modifications, or give suggestions?
Adrian: Directions.
Lou: Then at most BCP, as it is not creating nor modifying protocol.
Ross (with no particular hat): Last Call will be the place to put comments.
Adrian: Should we push it like this and wait for IESG review?
Kireeti: We should make up our minds before, as wording is different depending on selected option, if this is directive or suggestive. But we have to choose.
Lou: I agree with Kireeti, Standard Track is really different from BCP. We should take a decision.
Adrian: You would like a decision before Last Call?
Lou: I do not see how it could happen differently.
Zafar: A question on testing - how much has been done?
Richard: Talk to the vendors and see if they can give you additional info than available in the draft (interops are under NDA).
Zafar: Can bundling be added in?
Adrian: How much interop has been done?
Richard: Draft has been here long enough, can't wait forever for each new feature.
Ross (as individual): If Standards Track, we should be very precise on how fields are used on sending and receiving.
Adrian: Send comments to Richard, last round before LC.
=======================================================

=======================================================
4. ASON Solutions

   Routing solution (Dimitri)
   - Status and plans
Slides

Adrian: There was a fair amount of discussions on the list. Was a conclusion reached?
Dimitri: Loop prevention addressed, wording will be addressed for next revision. IS-IS draft will be submitted after this meeting.
Acee Lindem: You have a few comments of mine. Text was not clear wrt. hierarchical RAs, whether it is an 1:1 or 1:N
association between parent and child. And depending whether it is a downward
or upward point of view, the advertisement will differ.
Dimitri: We had e-mail exchanges, have to check.
Adrian: Will OSPF WG explode if ccamp does this draft?
Acee: We won't explode, we should get more review in ospf, though a lot of ASON-speak. There is some overlap with other hierarchy work - though this is only for TE.
Dimitri: Please give feedback if other terminology clarification is needed.
Adrian: How many have read this document. (A good show of hands)
Adrian: Time to be a WG document? (some show of hands)
Adrian: We will take it to the list.
============================================

=======================================================
5. MPLS/GMPLS Migration (Kohei)
   - Status and plans
Slides

Zafar: Is this for non-PSC GMPLS or for PSC GMPLS?
Adrian: This is talking about migration, you can not migrate from PSC to non-PSC. Can you find a feature which is not applicable to PSC?
Zafar: Some differences in P&R.
Adrian: Not sure on your comments, you can comment on the list.
Dimitri: How would the chairs like to evolve towards solutions?
Adrian: There are too many solutions, some identified are more preferable than others. Solution stripping would be the next step.
Kireeti: This is not interworking PSC and non-PSC, it's MPLS and GMPLS for PSC. GMPLS is a superset of MPLS (for those who are confused).
Adrian: Agree with Kireeti, this draft is on migration, the next one is
on MPLS over GMPLS.
Zafar: Concerns about the draft's impact on MPLS.
Adrian: Should bring to the list.
=======================================================

=======================================================

6. MPLS/GMPLS Interworking (Kenji)
-   Discussion
-   Next Steps
Slides

Zafar: Is it dynamic or static (management plane) based triggering?
Kenji: Both.
Nurit: Where are the scenarios documented?
Kenji: Not described in the document.
Nurit: We should also say why it's needed.
Adrian: Think that it would be useful to add an appendix.
Peter: I think this is a good document, should add references. I have some scenarios to add.
Julien: Why not do an upgrade from MPLS TE to GMPLS?
Kenji: We don't want to upgrade, that is the basis of the draft.
Deborah: Should make any comments during the next few weeks on the list and do a respin.
==============================================

==============================================

7. VCAT/LCAS (Richard)
   - Changes to the draft
   - Ready for the WG?
Slides

Dan Li: Need to address more on internal node capabilities, discovery.
Richard: What do you want to discover for? We don't advertise internal node information in GMPLS. Better done through the NMS. Need requirements from carriers if information is not available through the NMS.
Stephen: There's been OIF discussion on who retains knowledge of diversity, does each connection need to know what it is diverse from?
Adrian: Isn't this the same as requirements for protection/restoration?
Stephen: issue is requesting co-routing.
Richard: GMPLS VCAT is not a service, not same level of information. You can use the association object. We'll put more information in on supporting diversity.
Gert: What if this is started in the middle?
Richard: non-VCAT nodes do not need to know about VCAT portion.
Gert: If Ethernet, how does it know this can be supported in the middle?
Adrian: How does an intermediate segment advertise topology - that's routing, this draft is on signaling.
Richard: Hoping that GELS would use this, write an AS on using this for GELS.
Dimitri: Is diversity requirement anything new?
Richard: Adopted RFC 3946 for co-routed LSPs.
Adrian: Based on interest, will ask on mailing list for approval as WG draft.
==============================================

===============================================================================
8. Multi-Layer Network (Dimitri)
    - Progress and plans
    - Issues to be resolved
Slides

Zafar: Agree with much of this, don't agree with hybrid mode operation, supporting different switching capabilities.
Dimitri: This is not data plane, this is control plane.
Zafar: We already support multiple ISC capability flooding.
Dimitri: We aren't saying that we need to change.
Adrian: You are asking if can do as two logical nodes in one physical node?
Dimitri: That's the simplex model.
Zafar: Will do offline.
Evaluation Draft:
John: What is a virtual TE link?
Dimitri: Any suitable term is welcome.
Adrian: Potential TE link?
John: That would be better - here, advertises that it could be used. Potential is better.
Zafar: Does this overlap with migration?
Adrian: There was overlap, should not be anymore.
Adrian: Requirements almost done, evaluation, also with comments, solution will respin, and then ask for WG agreement.
===============================================================================

=======================================================================
9. MIB Module for OSPF-TE and OSPF GMPLS (Tomohiro)
-   Progress and plans
Slides

Adrian: I did talk with the OSPF MIB authors. Acee, is this ok to be done in CCAMP?
Acee: TE MIB is in CCAMP already, will read this.
=======================================================================

=======================================================================
10. Constraints for CSPF Path Computation (Kenichi)
-   Discussion
Slides

Adrian: Who has read? (only a few hands)
Adrian: Please read.
=======================================================================

=========================================================================================
11. Conversion between Permanent Connections and Switched Connections (Diego)
-   Progress
-   Ready for the WG?
Slides

Deborah: Who has read this document? (quite a few hands)
Deborah: Ready for WG document? (Only a few hands)
Stephen: This very complex, should add more on the complexity (pre-conditions).
Dimitri: Is this a requirements document or solution?
Diego: This draft lists the requirements, we need to think on solutions.
Adrian: Stephen, we do assume you have a control plane.
Zafar: Is this for migration?
Adrian: Yes
Zafar: Why has the reverse (control plane to management plane conversion) also been considered?
Adrian: In case something goes wrong.
Adrian: We'll do a respin and then consider for WG document.
==========================================================================================

=======================================================================
12. Inter-domain Recovery (Tomonori)
-   Discussion
Slides

Zafar: Your SRLG numbers may vary for different providers.
Tomonori: If global, then no issues.
Dimitri: It is more complex if use private addressing, so need to consider addressing, need to look at more.
Adrian: This work is just started, and Tomonori would welcome help.
=======================================================================

=======================================================================
13. Multiple Nodes Graceful Restart (Dan)
-   Discussion
Slides

Zafar: There was a draft with multi-node restart, and we pulled out the section as we thought it was too complex to include. So think rehashing.
Dimitri: How do you know the node is in that scenario, procedures need to be scenario agnostic.
Adrian: This document doesn't define procedures, just defines scenarios, and how works with existing procedure.
Dimitri: I understood it was a way to reuse tools, my question is how can node know which scenario and how to behave.
Adrian: I didn't think it needs to know, draft is just explaining.
?: How does a node know its neighbor?
Dan: They can use procedures already described
?: Discovery doesn't work though addresses. We submitted a draft to MPLS WG on this, we will submit here also.
=======================================================================

=======================================================================
14. Data Channel Status - LMP (Dan)
-   Discussion
Slides

Zafar: Why are RSVP refreshes not sufficient?
Dan: Refreshes used for LSP, not for this.
Adrian: Discuss on list, time is up.
=======================================================================

=======================================================================
15. ARP over GMPLS (Zafar)
-   Discussion
Slides

(Did not discuss - out of time)
=======================================================================