IETF 89 - RTGWG Minutes Scribe: Acee Lindem - acee.lindem@ericsson.com WG Chairs: Alvaro Retana, Jeff Tantsura * WG Update Welcome new Chair: Jeff Tantsura Thanks to old Chair: Alia Atlas Welcome new AD: Alia Atlas Thanks to old AD: Stewart Bryant * Document Status - Alvaro Retana (see slides): * "Advanced Multipath Requirements" in RFC Editor's queue. * "Remote LFA FRR" and "Advanced Multipath Use Cases and Design Considerations" have completed WG Last Call. * Most active documents are being covered today Alia: IPFRR MIB written long ago. Need people to bring work on. They don't necessarily have to be MIB experts. * Advanced Multipath Update - Curtis Villamizar (See slides) - Thanks to Lou Berger for Routing Directorate Review on "Advanced Multipath Requirements" * LFA Manageability - Stephane Litkowski (see slides) - Asking for WG Last Call. Jeff: Asks all to review - especially Service Providers. Alvaro: Polls for those that have read and think it is ready. Will be taken to the list. * Remote LFA Node Protection and Manageability - Chris Bowers (see slides) - Asking for WG Adoption: Alvaro: How many have read and how many feel it is ready for adoption? It will taken to the list. * Heuristics for pruning PQ nodes - Chris Bowers (see slides) Alvaro: Will this be put in draft format? Chris: No - it will be in the meeting materials. Jeff: Service providers please look at the results. * Microloop prevention by introducing a local convergence delay - Pierre Francois (See slides) - Asks for WG adoption. Acee: Please go through link down scenario again. Pierre: Backup path is not LFA. Alia (No hats): Likes current version of the draft since it provides a simple solution. Also thinks the LAN case should be solved and is useful. Pierre: Yes, we should add LAN. Stephane: Not complete solution, are you happy that it doesn't address node protection? Alia: Do you care about node protection? We need to talk about it. Thinks we should get this work done now without node protection. Stephane: Implementation exist. Alvaro: Anyone think this should not be a WG document? * Topology Independent Fast Reroute using Segment Routing - Pierre Francois (See Slides) Stewart: Not following completely. You won't expect traffic from PLR. Pierre: Capacity planning will need to consider failure cases and PLR. Pierre: Would be interested in analyzing other SP topology. Chris: Is the protected traffic Segment Routed or SPF IP traffic? Pierre: It it traffic engineered. Chris: Two use cases: IP traffic and Engineering. Pierre: WG document in RTG WG or SRRING WG? Alia (AD Hat on): SPRING MPLS use case is to make forwarding decision based on top two layers. Pierre: No Alia: Need two cascaded lookups or two labels examined. Blowing up the forwarding state. SPRING architecture needs to be baked. Hannes Gredler: Will recursive lookups be required for node protection - Yes, but this exists today for L3VPNs. Have you considered using RSVP for interface protection? Common to forget to integrate with existing technology. Document should not make assumption of technology used to setup protection. Pierre: Add another section covering the RSVP use case? Hannes: Just generalize the terminology so that any technology can be used for LSP setup. * Dynamic Path Selection (DPS) Based on Application - Arun Arumuganainar (See Slides) Acee: We have multi-topology routing (MTR) to solve this problem. Arun: Overlay could be replaced with MTR as the underlying topology. This is more of a framework. Bhumip: Static or dyanmic profiles? Arun: Static profiles. Bhumip Khasnabish: Set of predetermined profiles? Arun: Yes. Alia: Interesting and useful. We need requirements and use cases in a separate draft to evaluate what types of solutions can satisfy them. Arun: Can document the use cases but would welcome inputs on new use cases. Alia: Two level of draft. Use cases and requirements. Jeff: Thanks for bringing this work to RTGWG. Agrees with Alia on separating the problem statement from the solution. Also requests to remove the vendor specific terminology from the draft. * Generic Fault-avoidance Routing Protocol for Data Center Networks - Bhumip Khasnabish (See slides) Acee: So you always look at Negative Routing Table (NRT) first? Bhumip: Yes Hannes: Draft is highly subjective. Data Centers are using existing protocols without problems. Bhumip: Please send comments to mailing list. Hannes: I can do this. Network convergence doesn't follow link state dynamics - Fast reroute exists. Alvaro: Fast Reroute is not only part of charter. It is to introduce new technologies. Hannes: Routing protocols uses concepts from EIGRP. Beware of IPR on mechanisms. Alia: One of the functions RTGWG is to introduce new ideas and bring them to routing. (Speaking as AD) Jakob Heitz: What happens when you add links? Bhumip: Will add to document. Alvaro: Please send comments to the list. Alvaro: Further comments? If not, we will conclude this meeting of Jeff's Working Group ;^)