PCE Working meeting minutes - IETF-70 - Vancouver Thanks to Dan King for having taken the minutes. WG Update ( see the slides) - IGP PCE Discovery are in RFC Editor queue - Three new WG IDs - Need to close on the Inter-AS PCECP Requirement. - Policy framework passed WG last call. - PCEP has been stable for several revisions (close on minor issues); look to last call PCEP in January followed by BRPC, XRO, Path-Key, OF. WG Update Comments: Lou> What is stopping us sending it to the IESG Adrian> Work. Lou> What is the ETA on the write-up? JP> January 08 General Comments: JP> Do not hesitate to use the new tool for uploading drafts. 3) Inter-AS Requirements for the Path Computation Element Communication Protocol (PCECP) - draft-ietf-pce-interas-pcecp-reqs-04.txt (Kenji/Raymond/Nabil - 10mn) [25] JP> Authors did not have the time to post the rev-04 before cut-off date. They mention that the ID should be posted very soon. We will follow with a WG LC. 4) Update on Path Computation Element (PCE) Communication Protocol (PCEP) draft-ietf-pce-pcep-09.txt (JL - 15mn) [40] FT & Marben Interop PCC/PCE Session Management Computation Requests No Interop problems encountered Comments: JP> What did you test? Fabien> Scope of testing was only the main draft. We tested most except load balancing and SVEC. JP> Did you test error conditions? Fabien> Yes, path request unsupported object. JP> As an FYI two succesfull other Interop testing events took place. Note that the interop have been been on all the PCEP features. 5) Encoding of Objective Functions in Path Computation Element (PCE) communication and discovery protocols draft-ietf-pce-of-01.txt (JL - 5mn) [45] JP> Note that moving detailed discovery capabilities from the IGP to PCEP is compliant with the RFC. We have to do this work. Adrian> Was OF including in Interop. JL> No Adrian> is OF included in your implementation ? JL> Yes. 6) Preserving Topology Confidentiality in Inter-Domain Path Computation and Signaling - draft-ietf-pce-path-key-01.txt (Adrian - 5mn) [50] No Comments 7) Update on PCE MIBs (Emile/Kiran - 10mn) [60] draft-ietf-pce-tc-mib-01.txt draft-ietf-pce-disc-mib-01.txt Comments: JP> None of the authors could attend. Still very little feedback from the WG on the list. Please provide feedback, we are getting close to last call PCEP. Especially from SPs. 8) Update on Inter-layer (Eiji - 15mn) [75] draft-ietf-pce-inter-layer-req draft-ietf-pce-inter-layer-frwk-05.txt draft-oki-pce-inter-layer-ext-00.txt Comments: JP> Show of hands for WG doc. Looks to be enough support. Let's take it to the mailing list. JP> We have the components. When you start to have multi-layer optimization you can have oscillations. Would it be wise to have a discussion about dynamic changes in the framework document? In order to promote such a model we should be very cautious to promote dynamic behavior. Indeed, this might be relatively light protocol extension but the model itself (dynamic inter-layer reoptimization) may lead to serious issues. Eiji> We will add dynamic section. JP> Ross this work is for inter-layer engineering with optimization at multiple layers. We should be careful, we have PCE experience at a single layer, we need to be cautious about the dynamic nature of multi-layer engineering. Ross> Yes, need to be careful JP> For instance, a PCE may decide to dynamically provision a new lambda in response to some level congestion at the packet layer, leading to traffic reroute, potentially crossing some low utilization thresholds that could themselves lead to decommisioning some lambda, thus the risk of oscillations. We need to have place holder in the framework to discuss. Ross> Maybe we need to look at damping. JP> We will take this offline and discuss on the mailing list. 9) PCC-PCE Communication Requirements for Point to Multipoint Multiprotocol Label Switching Traffic Engineering (MPLS-TE) draft-yasukawa-pce-p2mp-req-03.txt (Adrian - 10mn) [85] draft-yasukawa-pce-p2mp-app-01.txt Comments: JP> Is P2MP within PCE charter, personally I believe this in the charter as its path computation. Ross> I might want to think about this more, it seems P2MP may be more important to P2MP. Maybe we want to specifically mention this in the charter. JP> Are people wanting to implement ? JL> This is important. I have some feedback on applicability draft. We agree P2MP computation is more important than P2P. But you mention in your draft that it is unreasonable to perform P2MP computation at the head-end: there are powerful heuristics and it is not unreasonable to expect that the head end to compute paths. JP> Who is interested in P2MP? Some show of hands, please email the list and discuss P2MP. 10) WSON frame/Information updates (Greg/Young - 10mn) [95] http://www.ietf.org/internet-drafts/draft-bernstein-ccamp-wavelength-switched-02.txt http://www.ietf.org/internet-drafts/draft-bernstein-ccamp-wson-info-00.txt Comments: JP> PCE is computation engine. It's not a Network Management/Provisioning tool Greg> We wanted to do RWA, we need information and this needs to be handed onto PCE. JP> You may want to clarify the term "Assignment" here. JP> Did you get to have a chat with the routing WGs? Greg> Not yet Adrian> What Greg has done is the first pass. This will provide input to the Routing WGs if this mechanism is proposed Greg> I think the Framework draft will live in CCAMP. Adrian> This is up for CCAMP on Thursday. The consequences for PCE are how do we poke PCE for computing the path. Greg> Yes, that's the next draft. Adrian> WSON info draft is clearly informational. It does not have direct impact on protocol work on the PCE WG but it background that is needed. 11) PCE WSON routing and wavelength assignment (RWA) (Young/Greg - 10mn) [105] draft-lee-pce-wson-routing-wavelength-00.txt Comments: JP> What is the model? Young> We are not constrained by an offline or online PCE model but we are looking at more online computation and not an NMS system. JP> Need to be clear about the term Assignment. Adrian> If you know about RWA you know it’s a X. If you are a routing engineer you use Assignment in signaling context. Need to add a description of the context of Assignment in RWA. Young> Agree JP> back to my comment about inter-layer. The dynamicity issue is relevant here as well. Young> This is applicable to MLN. Adrian> PCE operates on data, that data can come from anyway. JP> That data could come from the IGP Young> Is this applicable to the WG charter? JP> Compared to P2MP its more of a stretch. We will follow up on this. JL> Interesting draft. The first scenario on Page 4 is clearly in the charter but the 2nd scenario is not in the scope. Young> This is more of a twp stage computation. Adrian> This is multiple algorithm. JP> This needs clarification. Tomo> Will we focus on WSON or will the x (this?) focus on non-packet LSPs? Adrian> You may get ask in CCAMP if WSON work is applicable in TDM networks? Greg> The answer is most likely yes. These are global labels and NP hard computational problems. Its wavelength computation as well as path computation. JL> You are defining two or three RP flags. We only have 32 RP flags. This is a very specific, we should use a generic object. Adrian> PCEP review, why are they so few RP bits? Tomo> This work focuses on WSON, the concept should be applicable to any label switch path computation. 12) The use of SVEC (Synchronization VECtor) list for Synchronized dependent path computations - draft-nishioka-pce-svec-list-01.txt (I. Nishioka - 5mn) [110] JP> Needs more feed-back from the WG. Should not be Standard track but Informational.