IETF 110 PCE WG Meeting Minutes
PCE Working Group Meeting
Wednesday, March 10, 2021
UTC 12:00-14:00 Session I (Prague UTC+1 13:00-15:00)
- Chairs: Dhruv Dhody, Julien Meuric
- Secretery: Hari (Hariharan Ananthakrishnan)
1.1 Administrivia, Agenda Bashing (chairs, 5 min)
1.2 WG Status (chairs, 10 min) [15/120]
1.3 State of WG I-Ds and next steps (chairs, 15 min) [30/120]
- [Chairs] Thank you Deborah! Welcome John!
- [Haomian Zheng] draft-ietf-pce-pcep-stateful-pce-gmpls - Document is ready for WG LC.
2. Segment Routing
2.1 SR Policy (Mike Koldychev, 10 min) [40/120]
- [Dhruv] We use fixed value "1" for ASSOC-ID. We should ask WG for feedback if they have any other suggestions.
- [Mike] We have the extended association-id which encodes the color & endpoint, there is 16 bit assoc-id value inside the association object where "1" is specified. Please let us know your suggestions.
- [Boris Khasanov] In the implementation status, you talk about XR proof of concept. Is this jxr version or engineering version?
- [Mike] Currently its an engineering release. If another vendor is interested in interop you could ask for engineering patch to enable the feature.
2.2 Algorithm in SID (Samuel Sidor, 10 min) [50/120]
- [Dhruv] Can we conclude that all the comments from last meeting have been resolved?
- [Samuel] Yes, they are covered.
- [Dhruv] We can take this to the list and ask for confirmation.
- [Samuel] Feel free to open any new comments in the mailing list
- [Dhruv] I do have one comment which we can take after the SID verification presentation.
2.3 SID Verification (Ran Chen, 5 min) [55/120]
- [Mike] Flag is defined at the level of ERO sub-object. I am wondering if that is the right place? Instead should it be applicable to the whole segment-list or candidate path?
- [Ran] We would like to align with PCE-SRv6 and BGP SR Policy, and the same is done here.
- [Mike] Agree.
- [Dhruv] IDR also defines this per-SID. In IANA section please use TDB. Avoid allocating it ourself.
- [Dhruv] Looking at the minutes from 109, idea of having the "V flag" based on the PCE-SRv6 update. I think the authors between the two documents decided that its better that we maintain it separately. Anything that is missed between BGP SR-Policy and PCEP, lets fix that once for all. Discuss with IDR authors and get alignment with PCEP and IDR WG authors.
- [Ran] Yes, we will fix and discuss with IDR authors
- [Samuel] There are some discussion about merging this with multi-path draft. We can potentially match some of these into other objects.
- [Dhruv] My request would keep merge as one part and the second would be a proper check with IDR documents. Lets do an exercise on the mailing list and make sure alignment is there.
2.4 Multipath ERO (Mike Koldychev, 10 min) [65/120]
- [Vishnu Pavan Beeram] (from chat) It is clear that we need a TLV to carry color. We seem to have a couple of documents that define the color TLV: (a) https://tools.ietf.org/html/draft-peng-pce-te-constraints-04#section-3.5 (b) https://datatracker.ietf.org/doc/html/draft-rajagopalan-pcep-rsvp-color-00#section-3 The latter is an RSVP centric document, but it doesn’t preclude the TLV to be used for other path-control types. The authors of the 2 documents can sync up and ensure that there is a single placeholder for the TLV.
- [Pavan] There are documents that define color TLV. Originally we thought EX-Assoc-ID is sufficient. We need single placeholder. Please sync up with authors of other drafts.
- [Dhruv] There are other solutions than TLV. We can discuss this and resolve on the list.
- [Dhurv] This is next in line for our WG adoption. Please handle alignment soon.
3. Stateful PCE & PCECC
3.1 Native IP (Boris Khasanov, 10 min) [75/120]
- [Dhruv] Add a section for "BGP Consideration" and coordinate with IDR WG. Security consideration section is empty. There are impacts to security here.
- [Dhruv] PPA - Peer Prefix Association, its overloaded with multiple association types. If you could find a name that does not use association, that would be good to avoid confusion.
3.2 Inter-Domain (Olivier Dugeon, 10 min) [85/120]
- [Hari] Why is LSP Extended Flag considered as a 'Con' ?
- [Olivier] LSP Extended Flag draft needs to be adopted/implemented first and thus an extra dependency.
- [Dhruv] Olivier, please post details of the 3rd option on the list. WG, please provide your feedback.
3.3 IFIT (Giuseppe, 10 min) [95/120]
- [Boris] Do you plan to add implementation/inter-op section later on ?
- [Giuseppe] Yes, we can consider. Its a good suggestion.
- [Dhruv] Thanks for keeping the document aligned with IDR.
- [Dhruv] Do you want to create an IANA registry for Enhanced Alternate Marking Flags ? or Fixed set of flags.
- [Giuseppe] We would like this to be fixed set of flags, as per the 2 data plane option.
- [Dhruv] For other sub-TLVs we are referring to IAOM drafts directly and for this flag we are defining them as fixed. Just something to think about and we can discuss on the list
3.4 Operational Clarification (Mike Koldychev, 5 min) [100/120]
- [Dhruv] One use-case would be that, we allow NAI only as an option. It can carry path without allocating SID and PCC could pick the SIDs. More investigation is needed.
- [Mike] That option is allowed in SR-ERO.
- [Dhruv] ERO is path that PCE selects for NAI and RRO is the actual labels are represented.
- [Andrew Stone] interesting use case of RRO. The text doesnt prohibit SR-RRO.
- [Mike] In RSVP-TE, the RRO object has specific purpose like label recording and other use-cases. If we can find a meaning for it then its perfectly fine with me.
- [Pavan] RRO is used to reflect as actual path vs intended path (ERO). There could be instances where you push ERO and actual could be different.
- [Mike] This diverges from original RSVP RRO meaning and its fully optional.
- [Andrew] I have seen different vendors like Nokia report both ERO and RRO. Its a pointless data-structure that echo back. This is a observed behavior.
4.1 BIER-TE (Huaimo Chen, 10 min) [110/120]
- [Ran] we submitted BIER-TE draft in 2015 and its now in WG adoption queue. Please refer to the PCE Wiki and review it.
- [Huaimo] I referred your document in this draft.
- [Dhruv] Please build on the existing draft rather redefining. Currently PST and some of the things we are redefining.
4.2 PCEP-LS (Gyan Mishra, 10 min) [120/120]
- [Mike] From Cisco implementation, we dont have much interest. We already have BPG-LS which works. There has to be really good justification for doing this.
- [Dhruv] Does the use-cases which are highlighted here be of interest ?
- [Mike] I havent thought about those use-cases. Unless there is a clear case that something cannot be done with BGP-LS.
- [Dhruv] We should take this to the list.
5. If time permits
5.1 RSVP Color (Balaji Rajagopalan, 5 min) [125/120]
- [Dhruv] We dont have time to cover this.