PCE Meeting Minutes
74th IETF Meeting
Chairs: Adrian Farrel & JP Vasseur
Scribe: Daniel King
-------------------------------------------------
Tuesday: 1710 - 1810 Afternoon Session III
-------------------------------------------------
1) Administrivia (Chairs)
>JP Vasseur is MIA<
- Item 9 [draft-kumaki-pce-bgp-disco-attribute-03], has been dropped from the agenda as the presenter is not able to be present due to travel issues.
- Note from the chairs regarding WG presentations. Please be aware that they received many slide revisions at the last minute. Please try to prepare earlier and minimise revisions where possible.
2) WG status (Chairs) 5 minutes [10]
Status Page: http://tools.ietf.org/wg/pce/
- draft-ietf-pce-pcep
This work has become an RFC [RFC5440]. This draft was initially held up due to additional registry and security efforts.
- draft-ietf-pce-path-key
Was blocked due to PCEP [draft-ietf-pce-pcep] but can now move forward.
- draft-ietf-pce-pcep-xro
Was also waiting on PCEP, PATH-KEY [draft-ietf-pce-path-key] and BRPC [draft-ietf-pce-brpc].
>JP Vasseur Arrives<
-
draft-ietf-pce-of
Dependent on one discuss, see the mailing list poll (Is the lease load bandwidth
aiming at normalized bandwidth). JP received confirmation from a service
provider that the current definition was OK.
- draft-ietf-pce-global-concurrent-optimization
An updated has been posted recently and is currently being reviewed by the IESG.
- Two new WG documents, PCEP MIB [draft-ietf-pce-pcep-mib] and VPN Requirements [draft-ietf-pce-vpn-req]. Considering the status of PCEP the MIB document is very important.
- No questions regarding WG status.
3) PCEP Requirements and Extensions for WSON
Slides: http://www3.ietf.org/proceedings/09mar/slides/pce-6.ppt
Background reading: draft-lee-pce-wson-routing-wavelength-02.txt
Gregory Mirsky: I think that interoperability for impairments for RWA is not practical. Especially since meeting with the ITU and they are not requiring parameter interoperability. It's an over complication.
Young Lee: I was also at the Question 6 meeting, they mentioned that detailed simulation based validation may not be practical. They also discussed "flood leak" [G.698]. If the vendor does not want leak impairment data, then that should be separated from ITU-T. We can discuss this further.
Donald O’Connor: There was a meeting and that was a good first step. There needs to be further collaboration with the ITU. That process will take time. I would recommend the first step in PCE would be to do this without optical impairments.
Malcolm Betts: I was also at the joint ITU and CCAMP session and the very strong message was that - Yes we do have the concept of a black link and being able to attach third-party transmitters and receivers. But it is the owner of the black link that performs path computation - So there is not any notation of sharing impairment information outside of the context of the black link. I think we should keep RWA and impairments completely separated. While I am at the mike I would like to ask a question to the chairs. What's the relationship between this work in PCE and the WSON work in CCAMP?
Adrian Farrel: CCAMP has some framework documents; the decision was to partition the work into RWA with no impairments, impairments with some degree of complexity and impairments with a greater complexity. Those framework documents refer to PCE solving some of the computation problems. This document is intended to expand on those PCE discussions.
Julien Meuric: I would like to second Don's [Donald O'Connor] comment. Maybe a two-step approach is possible but both are required today. Second point, slide 4 [BER limit in PCEP request]. How would a PCE compute a route considering BER, when you can only measure BER once the LSP has been established.
Young Lee: We are proposing that you have an engineering rule. A policy server would apply a limit when you precision a lightpath. A PCC [as an NMS with an estimated BER] could also bound the BER when requested a path.
Julien Meuric: I do not really understand how this is possible.
JP Vasseur: Please continue this discussion on the mailing list.
4) Alternative Approaches to Traffic Engineering Database Creation and
Maintenance for Path Computation Elements
Presented by Greg Bernstein
Slides: http://www3.ietf.org/proceedings/09mar/slides/pce-7.ppt
Background reading: draft-lee-pce-ted-alternatives-01.txt
JP Vasseur: Who is interested in this work? [>6 show of hands] This topic has been silent on the mailing list. Please take this to the mailing list.
Greg Bernstein: We take this to the mailing list.
5) The use of SVEC (Synchronization VECtor) list for Synchronized
Presented by Itaru Nishhioka
Slides: http://www3.ietf.org/proceedings/09mar/slides/pce-1.ppt
Background reading: draft-ietf-pce-pcep-svec-list-01
JP Vasseur: Any comments or questions regarding this draft? [no comments]
Itaru Nishhioka: We will perform an final review, issue another update and request WG last call.
JP Vasseur: You will last call after Stockholm?
Itaru Nishhioka: Yes.
6) Status on Extensions to the Path Computation Element
Communication Protocol (PCEP) for Point-to-Multipoint Traffic
Engineering Label Switched Paths (Dan King - 5mn) [30]
draft-ietf-pce-pcep-p2mp-extensions-02.txt
JP Vasseur: Each time you fragment you must define the process [in the draft].
Daniel King: Yes.
7) PCE-based Computation Procedure To Compute Shortest Constrained P2MP Inter-domain Traffic Engineering Label Switched
Presenter Quintin Zhao
Slides:
Background material: draft-pce-pcep-inter-domain-p2mp-procedures-00.txt (Dan King - 10mn)
Nicolas Neate: I think that end-to-end optimization is not a requirement [in this draft].
Quintin Zhao: We believe that the inter-domain P2MP LSP path with that requirement is not scalable in some scenarios.
Nicolas Neate: It’s the case that it's not a tractable problem and you are working on the inter-domain optimization instead?
Quintin Zhao: Yes.
Daniel King: [end-to-end optimization] is an NP-hard problem. We are looking to provide a good solution, rather than the best solution.
JP Vasseur: This is an area that requires time to understand the proper dynamics. Creating extensions is not a hard problem. You may also want to combine mechanisms. Are you planning provide any data or simulations to understand the dynamics.
Daniel King: Yes. I think that’s a really good idea [modeling]. Having the data to sanity check the procedures and performing modeling to look at examples is good.
8) Requirements for PCE applied in Time-Division Multiplexing (TDM) Networks
Adrian Farrel presenting on behalf of Fatai Zhang
Slides: http://www3.ietf.org/proceedings/09mar/slides/pce-4.ppt
Background reading: draft-xx-PCE-reqs-for-TDM-00.txt
- Please review the draft and send questions to the mailing list.
9) BGP protocol extensions for Path Computation Element (PCE) Discovery in a BGP/MPLS IP-VPN
Slides: http://www3.ietf.org/proceedings/09mar/slides/pce-5.ppt
Background reading: draft-kumaki-pce-bgp-disco-attribute-03
- Kenji was unable to attend and the document was presented by a colleague.
Adrian Farrel: One task for this work, set in Dublin, was to make sure this document was in synch with the requirements document [draft-ietf-pce-vpn-req-00]. The VPN requirements document is now a WG document and Kenji has done a sanity check and assures me that this work is in step and that there are no new requirements. The authors of the requirements draft [Adrian Farrel and Seisho Yasukawa] need to agree.
End of meeting.
>JP Vasseur Leaves<