PCE Working Group Meeting - Monday, November 13, 2017; 15:50-17:20 Monday Afternoon session II o Slides: https://datatracker.ietf.org/meeting/100/session/pce o Etherpad: http://etherpad.tools.ietf.org:9000/p/notes-ietf-100- pce?useMonospaceFont=true o Meetecho: http://www.meetecho.com/ietf100/pce o Chairs: Jonathan Hardwick, Julien Meuric (remote) o Secretery: Dhruv Dhody 1. Introduction 1.1. Administrivia, Agenda Bashing (chairs, 5 min) - No comments on agenda bashing 1.2. WG Status (chairs, 20 min) [25/90] Jonathan Hardwick: Stateful PCE is published as RFC 8231 finally!! Good to see PCEPS published as well. Jon: For draft-ietf-pce-inter-area-as-applicability, re-check if we should include stateful PCE details now that stateful PCE RFC is published, and if so, the draft will go back to the WG. Dhruv Dhody: For draft-ietf-pce-pcep-yang, YANG model draft is now stable. Reviews from YANG doctors would be nice, and can go along with other TE YANG models. Dhruv: ACTN-related drafts (draft-ietf-pce-stateful-hpce, draft- ietf-pce-applicability-actn) are now aligned with the latest ACTN framework and stable. Stephane Litkowski: Association diversity draft (draft-ietf-pce- association-diversity) has addressed a few comments, a small change is to be exptected. For policy (draft-ietf-pce-association- policy), we are considering adding another TLV for variables for policy. Dhruv: Regarding scheduling (draft-ietf-pce-stateful-pce-lsp- scheduling), an update was made to keep it aligned with TEAS documents as well as RFC8231 and other PCE WG document. More reviews would be nice. Zafar Ali (proxy by Kamran): For draft-ietf-pce-pcep-stateful-pce- gmpls and draft-ietf-pce-remote-initiated-gmpls-lsp, will be refreshed tonight. Haomian offered to take over editor role for draft-ietf-pce- enhanced-errors. 2. WG's I-Ds 2.1. Stateful PCE for P2MP LSPs (Pavan, 5 min) [30/90] draft-ietf-pce-stateful-pce-p2mp-05 Vishnu Pavan Beeram: Request WG LC Jon: Request a few reviews of the document as we move this document to WG LC and get more eyes on it 3. New I-Ds 3.1. PCEP Extensions for SR leveraging the IPv6 data plane (Dhruv Dhody, 10 min) [40/90] draft-negi-pce-segment-routing- ipv6-00 Jeff Tantsura: This work is useful and makes sense. Jon: It feels obvious to do this. Heads up that LSP-setup-type draft will change and that would need to be updated here. This draft has a dependency on an issue in the Segment routing draft and that should be discussed on the list. Jeff: Regarding the dependencies with SRv6 network programming work, keep the your extension simple, if you include SRv6 network programming details, you would need to wait for a while. Dhruv: Yes, we could keep it simple, and make sure it can be extend easily later. 4. Previously Discussed Topics 4.1 Association for Path Protection (Dhruv Dhody, 10 min) [50/90] draft-ananthakrishnan-pce-stateful- path-protection-04 draft-ietf-pce-association-group-04 Jon: Association extended to support Protection would be a good idea. Poll: How many have read this document? (around 10, good number!) Jon: Would take it to the list. 4.2 Association for Bidirectional LSP (Rakesh Gandhi, 10 min) [60/90] draft-barth-pce-association-bidir-03 Poll: How many have read this document? (around 6) Jon: Would take it to the list, make sure enough eyes on it. Jon: Clarification question on double-sided bi-directional lsp. Himanshu Shah: Interesting draft, would like to see this adopted. 4.3 PCEP to support Resource Sharing (Haomian Zheng, 10 min) [70/90] draft-zhang-pce-resource-sharing-05 Himanshu: Clarification question on diversity v/s sharing Poll: How many have read this document? (around 10-12) Poll: How many think it is a reasonable base and dopted? (about the same) 4.4 Syncronization between Stateful PCEs (Dhruv Dhody, 5 min) [75/90] draft-litkowski-pce-state-sync-02 Adrian Farrel: To confirm, are you assuming the sync is always from a master PCE to slave PCE? And never to a master. Dhruv: Yes, the forwarding rules are defined in such a way to make sure that the master has the latest information. Jon: Wants to understand the use case better. Are there multiple active PCEs in the network for redundancy and load balancing? Dhruv: In the -00 version, only computation loop and redundancy was considered, it was further expanded to include more use-cases like inter-domain and H-PCE as well. Jon: Could these use cases be addressed by horizontal scaling within the PCE software? Do we need a communication standard protocol between different PCE instances? Dhruv: Both are needed. In some case PCE instances inside cluster could use DB sync techniques. Jon: You have some other use case on inter-domain, not sure the use there either? Dhruv: For inter-domain LSP, we need to exchange information for inter-domain LSP crossing multiple domains between cooperating PCEs for stateful PCE model and similarly for H-PCE keeping the database synchronized from child to parent. Poll: How many have read this draft? (only 3-4 hands raised) Jon: Not enough traction and get more people involved. Stephane: The application based redundancy (such as DB sync) have some constraints, such as all PCE instances must be in a single site, which is not acceptable. We require a more robust handling and using PCEP is helpful and simple. Jon: You think it is not possible with DB sync with a back-end DB. Stephane: Theoretically yes, but practically there are limitation and clustering solution works well at same site and has 10-15 ms latency which cannot be made to work across multiple sites.