CCAMP WG Minutes - 59th IETF Seoul

["Adrian Farrel" <[e-mail]>]

Common Control and Measurement Plane WG (ccamp)

THURSDAY, March 4 at 0900-1130
===============================

CHAIRS: Kireeti Kompella <[e-mail]>
        Adrian Farrel <[e-mail]>

===
Group Admin
---
Chairs
  Admin - Nothing much to say (in English anyway)
        - In Korean, however, the following was said:
          "Jigeumbuteo CCAMP meetingeul sijakhagesumnida".

  Agenda bash (5 mins) - No changes
  Status of WG drafts and milestones
    Adrian's slides showed that we do have some draft
    congestion in the WG.
    - RFC editor queue
    - status of IANA for SONET/SDH
      Kireeti talked about an issue with SONET/SDH IANA
      assignments
    - need a means to get early assignments.
      There is WIP to accomplish this, and it is moving
      ahead.
    - future allocation of "experimental" values

Liaisons
---
Marco Carugi talked about work in SG-13 (SG13 liaison).
  He covered topics, new study areas, timescales, objectives
  and status. They are also looking for people interested in
  doing work in these areas.

  An L1 VPN questionnaire and framework draft were attached
  to the liaison.

  Tomonori Takeda talked about the technical issues and
  details of the work.

  Monique Morrow had a couple of clarification for Marco -
  When will the consent portion of the work be done in the
  ITU?

    Marco said that the different pieces of work were
    progressing at different speeds. Some material is
    already embodied in recommendations. The next SG13
    meetings are in June and September.

  Dimitri Papadimitriou asked if the draft (l1vpn
  framework) provided in the liaison could include a
  summarization (as conclusion) on the expected GMPLS work
  for the CCAMP WG, this would clarify the intent of the
  liaison in term of expected effort for the CCAMP WG

    Kireeti answered. If CCAMP's rechartering this month
    results in the addition of L1VPNs to the charter, then
    a Liaison response from the IETF will include this
    information, plus a request for a cooperative effort,
    preferably along the lines of the ASON routing work,
    wherein the ITU-T defines the requirements and the IETF
    does the protocol extensions.

  Alex Zinin said that we will have to make a decision at
  some point as to whether or not we want to do this work
  here.

  Kohei Shiomoto said that the protocol for the L1VPN should
  be developed at the IETF as long as it uses IP protocol.
  There are already internet-drafts on GVPN and CCAMP is the
  best place to discuss it.

  Deborah Brungard said that there is work and some synergy
  and that we should continue to work on this.

    Monique Morrow agrees that we should work on that.

    Marco added some comments that were not captured in the
    minutes.

    Malcolm Betts said he also feels that we should do this.

  Adrian took a quick poll and it seems as if nobody is
  against doing this work.

  Kireeti reminded people to continue this discussion on
  the list.

---
Lyndon Ong talked about work in SG-15 (3 liaisons).

  Liaisons were on ASON routing requirements, response to
  comments on Q14 for G.7713.2 and comments on the CCAMP
  ASON signaling requirements draft.

  Lyndon spent much of the time on the details of response
  to comments on Q14. It seems that some of the differences
  in architectural models revolve around "end-to-end" and
  "call segment" operating models.

  Kireeti asked for the reply by date.

    Lyndon did not have that.

    Steve Trowbridge said that the meeting starts on April
    19th

  Dimitri had a question on the deadline. There is a
  deadline on G.7713 (April 2004), isn't there a similar
  deadline on G.7713.2 (since this is the document to which
  convergence is expected) ?

    Lyndon said that he had not gone into that. He gave a
    reason, but this was not captured in the minutes.

  Deborah said that the liaison for 7713.2 does not say any
  thing about convergence.

    Lyndon said that they are still looking for a "meeting
    of the minds".

  Deborah said that there is an issue with G.7713.2 because
  of compatibility.

    Lyndon said that yes there has been a lot of discussion
    of compatibility questions and requirements.

  Kireeti said that we should not discuss this here.

  Steve Trowbridge added some comments that were not
  captured in the minutes.

  Kireeti asked the WG to take this discussion to the list
  and try to keep that discussion on a productive basis.

  Adrian said that he wanted to recognize the efforts of
  the ITU folks in this work.

===
ASON Requirements and Solutions
---
Dimitri Papadimitriou presented status of ASON Signaling
Requirements (draft-ietf-ccamp-gmpls-ason-reqts-05.txt.

  The requirements were driven by last year's liaison from
  the ITU.

  The discussion slides proposed to defer to the GROW WG
  (cf. RIFT WG item) concerning the (external) non-IP
  reachability issue since much broader than just CCAMP
  GMPLS/ASON context

  After this meeting, Dimitri would like to re-spin the
  draft and have a two week last call.

  Lyndon said he want to capture the requirement on "non-IP
  reachability" - whether or not we will work on it here

  Kireeti said that we first need to understand importance
  of this and then we can look to the ADs for guidance on
  handling this.  He also said that we should take some time
  to work out what we want to say to the ITU when we include
  the current draft.

---
Dimitri Papadimitriou gave status ASON Signaling Solutions
(draft-ietf-ccamp-gmpls-rsvp-te-ason-01.txt) status.

  He would like feedback on whether or not the current draft
  deals correctly with the session attribute object that
  encodes the long call_id (alternatives were also proposed)

  His objective at this point is to try to have a document
  ready for last call for the next IETF 60 meeting in San
  Diego

  Lyndon suggested that we might remove the comparison with
  G.7713 from the draft.

    Adrian asked if this meant that the interworking draft
    for RFC3473/4 interworking was now obsolete.

      Lyndon said maybe, if interworking is removed as a
      requirement.

---
Lou Berger talked about Egress Control -
draft-berger-gmpls-egress-control-01.txt -

  Original egress label control became explicit label
  control. This draft attempts to capture the original
  intent.

  He wants to know if the WG feels that this is ready to
  be a BCP and what the chairs think the next steps should
  be.

  Lou re-iterated that the purpose and scope of the draft
  is for clarification. He does not see any value in adding
  to this intent or combining it with other work.

  Adrian then took a poll and nobody objected to take this
  on as a WG item (more than a third were in favor).

---
Lyndon Ong went over status on ASON Routing Requirements -
draft-ietf-ccamp-gmpls-ason-routing-reqts-02.txt

  He includes in his presentation the Design Team's
  conclusions as to what there is agreement about what's
  missing from GMPLS (delta), and what are the areas on
  which there is no agreement about what's missing from
  GMPLS.

  Vishal Sharma asked if the three issues (slide 6) were
  already opened up for discussion on the list, or would
  they be formally opened up with the DT members initiating
  a discussion on these on the list?

    Lyndon Ong replied that a discussion had not been
    formally opened up yet (although people were free to
    discuss/comment).

      Kireeti asked Lyndon to more formally open this
      discussion on the mailing list.

  Vishal Sharma said that he supports this.

  Kireeti said he would like - after checking with the AD -
  that we should take this work to the IS-IS and OSPF WGs.

    Alex Zinin said this is a good idea.

===
Tunnel Trace
---
Ron Bonica presented status on draft-bonica-tunproto-05.txt

  The solution is very similar to Trace-Route but does not
  require that each node in a tunnel supports TTL decrement.

  He gave a few examples as to how the idea in the draft
  will work in a few scenarios.

  There are a couple of outstanding issues:
  - trace requires a route to tunnel head end
  - integration with LSP ping.

  He would like to get the draft accepted as a WG draft.

  Yakov asked what SPs use today for tunnel tracing.

    Ron said that in some case people can use ICMP for MPLS.

  Yakov then asked if we could get a BCP on what people are
  doing.

    Ron asked if he should resubmit his earlier draft on
    this.

      Kireeti said that we do not want to decide that now.

===
Protection and Restoration
---
Dimitri Papadimitriou presented status on the work of the
Protection and Restoration Team - specifically:
1) draft-ietf-ccamp-gmpls-recovery-analysis-02.txt
2) draft-ietf-ccamp-gmpls-recovery-functional-01.txt
3) draft-ietf-ccamp-gmpls-recovery-terminology-03.txt
4) draft-lang-ccamp-gmpls-recovery-e2e-signaling-02.txt

  He gave estimates on the timing for each of the above
  drafts (estimated completion dates).

  He outlined the changes to the e2e signaling ID (draft 4,
  above).

  He encouraged the WG to really read the documents and
  comment.

  Kireeti polled for consensus on the following:

    a) Analysis - last call? Some support, no objection
    b) Functional - last call? Some support, no objection
    c) Terminology - last call? Some support, no objection
    d) e2e Signaling - WG document? Some support, no object

   Kohei Shiomoto said that the e2e Signaling draft does not
   address the misconnection issues raised in the mailing
   list.

     Dimitri answered that it is addressed in 8.3 of the
     draft.

       Kohei said that the misconnection issue does not
       happen only in the P&R context but also in more
       general context and therefore should be addressed
       in more general context as well.

         Kireeti said that the question should be continued
         to the mailing list.

  People at the microphone were asked to take their
  questions to the list.

---
Lou Berger presented an overview of work on Segment
Recovery - draft-berger-ccamp-gmpls-segment-recovery-00.txt

  He also talked about what still needs to be done (next
  steps), including more usage scenarios, more explanatory
  text and see if the WG will adopt this work.

  Arthi Ayyangar asked if the association object is required
  even if we are only doing segment recovery (as opposed to
  e2e).

  Arthi asked why couldn't we extend the Detour Object to
  achieve the same result. Kireeti asked her to take to the
  list.

  Richard Rabbat asked if this draft raised the same issues
  as the e2e signaling draft in terms of misconnection.

    Kireeti replied that they did not know if there were
    misconnection problems.

      Richard asked that the discussion about misconnections
      be moved to the mailing list in the interest of time.

  Adrian polled for support of accepting this as a WG draft.
  There was moderate support and no objection.

===
Inter-Area/AS
---
Arthi Ayyangar talked about the status of the merged draft
on Inter-area/AS signaling -
draft-vasseur-ccamp-inter-area-as-te-00.txt

  The draft currently represents a full merge - work is
  still required to strip out redundant and unneeded text.

  She said that the authors encourage people to come forward
  with their comments.  She would also like to see if there
  is interest in this work becoming a WG document.

  Vishal Sharma said that the work should apply to general
  path computation domains and GMPLS LSPs.
  In response to Arthi's question on Slide "Outstanding
  Issues" (about whether detailed description of various
  path computation algorithms should be part of this
  document or separate document(s)), he supported the
  document being split into a framework document, discussing
  signaling, and another document(s), discussing the path
  computation mechanisms, since the latter do not need to be
  standardized.
  In response to Slide "Outstanding Issues: Size of the
  document" and for clarity, he supported the splitting of
  the applicability statement into an independent document.

  Dimitri agreed on the subject of separating the document.
  In addition, he questioned about the relevance of using
  the LSP_Attributes to signal the signaling method for the
  intra-area/-AS provisioning of the LSP.
  In particular, he proposed to not include protocol
  procedures within examples/scenarios that makes the
  document difficult to read.

    Arthi asked that Dimitri take his specific comments to
    the list.

  Kireeti said that he agrees that the document needs to be
  split - one as a signaling and another (informational) to
  provide examples for path computation. He also said that
  we need a separate applicability document.

    Vishal Sharma then said that he would be happy to help
    with these tasks.

---
Vishal Sharma talked about work on Inter-area path
protection
draft-dachille-inter-area-path-protection-00.txt

  He provided a brief overview of how it works, and showed
  how it relates to other work in progress. He also listed
  the next steps.

  He emphasized that this is really a generic mechanism for
  diverse path computation, and protection is one
  application of it, so the authors would respin with a new
  name and emphasis to reflect this."

  Zafar Ali asked how this would work if there is a failure
  at the time during which the backup path is being setup.

    Vishal replied that the solutions to this were, so far,
    not discussed in the draft, but that there are several
    options.

    He then outlined some of the options. E.g. either
    default in such a case to a sequential computation, and
    use XRO to exclude the link/node where backup path setup
    failed, and retry the backup (and optimize both primary
    and secondary later using the techniques in the draft).
    Or, set up the primary and the backup again, using the
    techniques described in the draft.

    Vishal said they would be happy to add some discussion
    in the document, and welcomed feedback on the list.

  Zafar asked how this work relates to PCS/PCE work.

    Vishal replied that it could actually be made use of by
    the PCS/PCE approach, and could be viewed as
    complementary.

  Kireeti asked that further discussion be taken to the
  list.

  Vishal said he welcomed further feedback on the document.

  Dimitri asked why, knowing that the proposed approach
  works as expected in the intra-domain case when the
  number of ABRs (where computation can be executed at each
  stage) does not increase, this approach is so focused on
  optimization (since it can't be achieved if this
  condition is not met).

    Vishal clarified that the focus of the work is to
    propose a generic mechanism to facilitate diverse path
    setup by communicating alternate path info, with
    optimization a desired goal (for reasons explained in
    the document).

    Vishal added that given the network model (where border
    nodes are not assumed to have visibility in areas other
    than their own), the scheme was not trying to be
    globally optimal.

    Vishal explained that in such cases some selection needs
    to be performed at each stage.

  Kireeti asked that further discussion on this should be
  taken to the list.

  Also, he said that Dimitri had a good point - we need to
  define criteria on which any optimization is based.

  Kireeti concluded by saying that path protection and
  inter-area are both in the charter, but that this document
  could only be considered for a WG document after there was
  discussion about the document on the list.

===
Control Pane Resilience, Hello Protocol and Graceful Restart
---
Young Hwa Kim gave a presentation on Requirements for the
Resilience of Control Plane in GMPLS -
draft-kim-ccamp-cpr-reqts-00.txt

  He described the reasons why control plane resilience is
  needed.

  Zafar asked how control plane resilience is different from
  anything else in IP.

  Steve Trowbridge said that their is also some work in this
  area in the ITU and he would try to get this in as a
  liaison as soon as possible.

  Kireeti said that this is an important discussion and
  there are a lot of things to do. Specific topics should be
  raised on the list when appropriate.

---
Lou Berger went over Extensions to GMPLS RSVP Graceful
Restart
draft-aruns-ccamp-rsvp-restart-ext-00

  He emphasized that egress restart is already covered in
  RFC3473 and this work has no effect on that functionality.
  He gave a brief overview and listed open issues.

  Next steps include merging with other restart drafts and
  seeing if this work can become a WG draft.

  Arthi Ayyangar said that the text at the beginning of the
  draft seems to talk only about the recovery ERO, although
  using the RecoveryPath one can recover many objects
  besides the ERO. So the text should be clarified to this
  effect.

    Lou asked if she would like to contribute text.

  There was a discussion on adjacent node restart.

  Arthi asked why adjacent node restart was an issue being
  addressed in RSVP-TE. She pointed out that before this
  becomes an issue to be solved in RSVP-TE, we would first
  need to check if adjacent node restart even works for
  IGPs.

  The chairs then asked for other discussion to go to the
  list.

---
Zafar Ali talked about Extensions to GMPLS RSVP Graceful
Restart
draft-rahman-ccamp-rsvp-restart-extensions-00.txt

  Kireeti said that he appreciated the honesty of the
  authors in acknowledging other work.

  Nurit Sprecher asked about the relationship to FRR and
  similar issues.

    Adrian agreed that these were important issues and had
    been raised on the list in recent days. He asked the
    authors to make sure that they cover the points in the
    draft.

---
Zafar then covered modifications to Hello procedures
1) draft-ali-ccamp-rsvp-node-id-based-hello-00.txt
2) draft-ali-ccamp-rsvp-hello-gr-admin-00.txt

  He wants to go forward with draft 1 above.

  Adrian polled and there was some interest and no strong
  objection.

  Kireeti said that this work cannot be informational if
  it has - or proposes - changes to a standard.

  Zafar also wants draft 2 to be a WG document.

  Kireeti said that we need to take this to the list, but
  Zafar also needs to socialize the work he is doing so that
  people may decide whether or not this is work we want to
  do.

===
Everything Else
---
Emmanuel Dotaro gave status of Multi-region protection -
draft-vigoureux-shiomoto-ccamp-gmpls-mrn-04.txt

  He briefly covered changes since previous versions.

  He proposes that we may need to make changes to the
  charter to include all of this work. In particular to
  include in the charter the complete set of GMPLS
  mechanisms for integrated control planes, and not only
  multi-layer recovery (as it stands today).

  Adrian suggested that the authors need to get more people
  involved in this important work and revisit this later.

---
Jean-Louis Le Roux - Advertizing TE Capabilities in IGPs
draft-vasseur-ccamp-isis-te-caps-00.txt

  He would like to have this accepted as a WG document.

  Adrian asked to hold off on this until after the OSPF talk
  below.

---
Seisho Yasukawa
draft-vasseur-ccamp-ospf-te-caps-00.txt

  He would like to have this accepted as a WG document.

  Regarding both drafts, Kireeti is not sure that this work
  belongs in this WG. The decision is driven by the
  generality of its applicability. If we do take it on, their
  needs to be a functional specification (independent of IGP)
  as well.

  He asked that further discussion be taken to the list.

---
The Following presentations were postponed as we ran out
of time. Adrian made a couple of brief comments as follows:
  ---
  Zafar Ali - Explicit Resource Control and Tracking
  draft-zamfir-explicit-resource-control-bundle-03.txt

    This work concerns identification of component links in
    EROs and RROs.

    A small group is currently examining other issues
    concerning identification of component links in all
    aspects of GMPLS. A draft is expected soon. Please mail
    Adrian or the list, if you want to be involved in this
    work.

  ---
  Lou Berger - Alarm Reporting
  draft-berger-ccamp-gmpls-alarm-spec-01.txt

    This draft is stable and complete in the view of the
    authors.

    A quick poll showed some support for this being a WG
    document, and no opposition. This will be taken to the
    list.