Minutes IETF104: pce
minutes-104-pce-02
Meeting Minutes | Path Computation Element (pce) WG | |
---|---|---|
Date and time | 2019-03-28 08:00 | |
Title | Minutes IETF104: pce | |
State | Active | |
Other versions | plain text | |
Last updated | 2019-04-01 |
minutes-104-pce-02
PCE Working Group Meeting – 9:00-10:30 28th March 2019 Thursday Morning session I o Chairs: Julien Meuric, Dhruv Dhody, Adrian Farrel o Note Taker: Hari (Hariharan Ananthakrishnan) o Slide: https://datatracker.ietf.org/meeting/104/session/pce o Etherpad: https://etherpad.ietf.org/p/notes-ietf-104-pce?useMonospaceFont=true o Meetecho: http://www.meetecho.com/ietf104/pce 1. Introduction 1.1. Administrivia, Agenda Bashing (chairs, 5 min) [Julien Meuric]: Try to make use of the Wiki. [Julien]: Will start discussion on the Implementation Policy in the Mailing List. 1.2. WG Status (chairs, 10 min) [15/90] [Dhruv Dhody]: No new RFC since Bangkok. WG document status is up to date in Wiki. 2. WG I-Ds 2.1. Stateful PCE for GMPLS (Young Lee, 5 min) [20/90] draft-ietf-pce-pcep-stateful-pce-gmpls-10 [Young Lee]: Merged draft-ietf-pce-remote-initiated-gmpls-lsp to draft-ietf-pce-pcep-stateful-pce-gmpls [Dhruv]: Thanks for the merge, do another editorial pass and post an update. Higher preference will be given in the queue. Lookout for the comments to GMPLS extension that is with the IESG. 2.2. Enhanced Errors (Zheng Haomian, 10 min) [30/90] draft-ietf-pce-enhanced-errors-05 [Dhruv]: Clarification Question - What is on-demand? [Haomian Zheng]: Unspecified. Can be 0 or 1 depending on requirements. [Adrian Farrel]: Are you expecting authors to do this? or will you be talking to the authors and help them. [Haomian]: Yes, we will talk to authors offline. [Dhruv]: Lets provide the TEXT to the authors in the mailing list. Make sure the target users of this I-D actually uses this mechanism. [Deborah Brungard]: Update wiki with author guidelines and it will be useful, can also add security considerations. [Adrian]: Similar to security considerations section, a section on error handling in multi-PCE scenarios could be considered for informational documents as well. 3. PCE as a Central Controller 3.1. Native IP (Aijun Wang, 5 min) [35/90] draft-ietf-pce-pcep-extension-native-ip-03 [Adrian]: Has this been circulated to IDR WG for information. Send email to IDR mailing list, since this is related to BGP work. [Dhruv]: In TEAS the related documents are experimental, we have this in standards track in PCE? [Pavan Beeram]: The two docs we have in TEAS is experimental and no plans to move to standard. [Adrian]: Why are they in experimental? [Pavan]: Lack of broad interest but a few interested folks. [Aijun Wang]: Can this be changed later? [Adrian]: Yes [Pavan]: I don't see why the PCE Draft to be in standards track. [Dhruv]: I prefer to see this in the experimental track. We can take to the list. [Dhruv]: Thanks for reusing PCECC objects instead of creating new ones. We need more description and clarity in the document with respect to the three TLVs. Explain the procedure and explain the interaction between PCE and BGP. [Deborah]: You can go publish as experimental and come back and take it to standards track. Focus on maturity and implementation. [Adrain]: TEAS doc is more of framework and choice was between Informational and experimental. Here it is protocol and choice is between standard and experimental. So this is about implementation status. We can make the decision when we go to AD while requesting RFC publication. 3.2. PCECC Extensions (Mahendra Negi (remote), 5 min) [40/90] draft-ietf-pce-pcep-extension-for-pce-controller-01 draft-zhao-pce-pcep-extension-pce-controller-sr-04 [Mahendra Negi]: Request for Adoption for PCECC-SR draft. [Tarek Saad]: There could be label allocation, are you thinking of carving label space to resolve conflicting label assignment between PCC and PCECC. [Dhruv]: Yes 3.3. PCE Controlled ID space (Aijun Wang, 10 min) [50/90] draft-li-pce-controlled-id-space-02 [Julien]: Is this 'experimental' on purpose ? [Cheng Li]: We chose experimental because it might be easier; we can opt for Standards track. 4. Segment Routing 4.1. SR Policy (Mike Koldychev (remote), 10 min) [60/90] draft-barth-pce-segment-routing-policy-cp-02 [Jeff Tantsura]: Policy and LSP should not be used interchangeably. In the document the term SR LSP is used. [Mike Koldychev]: SR Policy = PCEP association. We will correct it and use PCEP LSP term (instead of SR LSP). [Jeff]: How about use of Binding SID when the LSP/Policy cannot be instantiated end-to-end? [Mike]: Binding SID should work as before. Still belongs to the PCEP LSP (candidate path) [Jeff]: The endpoints are not the same as where the binding SID is expanded, if you cannot push the label, you might use binding SID to segment. [Dhruv]: This document is not talking about how to compute the candidate path, but how to group them. [Jeff]: It should interwork. [Dhruv]: Yes, also the current document assumes only one SID list in the candidate path which is different from what the SR policy draft says. [Dhruv]: To authors working on the SR drafts, please make sure there is alignment between BGP and PCEP documents. [Mahendra]: Can PCE and PCC initiated candidate paths co-exist ? [Mike]: Yes, they can co-exist. Protocol-origin could be used. 4.2. Path Segment & Bi-directional SR (Cheng Li, 10 min) [70/90] draft-li-pce-sr-path-segment-04 draft-li-pce-sr-bidir-path-04 [Tarek]: About terminology, SR-PATH, LSP, Candidate path, Segment List. Could this be the segment list ID or LSP ID? [Cheng]: Path segment can be used for identification of SID List or SR Policy? [Tarek]: How can you identify one candidate path with many segment list? [Rakesh Gandhi]: It depends on the use-case. You can have path segment for the whole policy or segment list. [Tarek]: Expand on the document. [Dhurv]: Right place would be in the spring document. Spring WG has adopted. In PCE should we adopt this work? [Dhruv]: How do you want to move these document, both of them together? [Cheng]: Path segment is the base and key use case of path-segement is bi-dir. So they could be both adopted. [Adrian]: We will put them in pending queue for adoption in the wiki. 5. Other Topics 5.1. Stateful PCE Inter-domain (Julien Meuric, 10 min) [80/90] draft-dugeon-pce-stateful-interdomain-02 [Adrian]: Is there an assumption about the end domains are under the same admin control? [Julien]: There is none. BRPC is build on assumption that two domain might be under different control where as H-PCE is under the same. [Adrian]: Are you aware there is a document in BESS on domain gateway discovery? Similar, including discovering attachments. That is approaching last call. [Julien]: I am not aware. I will look into it. 5.2. PCEP for BIER-TE (Ran Chen, 10 min) [90/90] draft-chen-pce-bier-04 [Dhruv]: This was presented in the BIER WG. Talked to the BIER chairs. We will talk among ourselves and conclude the home for this. [Dhruv]: YANG work is going on in BIER WG. BIER YANG work could be the reference for us. Alignment should be done between YANG and PCEP work. [Jeff]: PCEP-BIER is exclusively for BIER TE. [Dhurv]: There is a confusion. There is BIER-YANG, BIER-TE YANG. This document is for BIER-TE but the PCECC extension could be for both BIER/BIER-TE. This needs to be clarified. [Lou Berger]: This comments is more for the BIER WG. Confusion in terminology; BIER policy and path steering (similar to spring)? We should be consistent. Adrian is helping us to update on what TE means in the IETF. Policy and Steering is what we are doing here. [Zheng (Sandy) Zhang]: BIER TE can be used without BIER. PCECC can be used for BFR-ID allocation for BIER and adj-id is for BIER-TE. This document is for path steering for BIER-TE. [Dhruv]: We are in agreement. ==Meeting Adjourned==