Routing Area Working Group WG Minutes WEDNESDAY, July 23, 2014 - 15:20-16:50 Afternoon Session II Chairs: Alvaro Retana, Jeff Tantsura Scribe: Acee Lindem WG Status Web Page: http://tools.ietf.org/wg/rtgwg/ 1) WG Status Update - Alvaro Retana, Jeff Tantsura (See slides) 1 New RFC 3 Drafts have completed WG last call 6 Active Drafts (LFA/RLFA Manageability ready for WG last call) 1 Draft (Microloop prevention) to be adopted by WG 2) IPFRR MIB - Stephane Litkowski (See slides) Jeff Tantsura - Thank you for taking work, Please review on the list. Alia Atlas - Should this extend the base IP Forwarding Table MIB (RFC 4292) rather than adding an instance? 3) SPF U-Loop Problem - Stephane Litkowski (See slides) Stewart Bryant: Are you going to cover the interface up case as well as down case? Alvaro: Hold your questions. 4) SPF Backoff Algorithm - Bruno Decraene (See slides) Peter Ashwood: Backward compatibility? Bruno: No - same considerations. Stephane: Uniform network is ok. Peter: Migration can be long. Stephane: Clearly you should control your network. Alvaro: SPF delay and backoff implemented already. I don't think standardizing value is what WG should do. Could be BCP. Stephone: Values are not standardized. Algorithm is standardized. Cengiz Alaettinoglu: Have seen these problems. Only occurs when there is high churn and there will still be uloops due to flooding differences. Bruno: Not going to eliminate uloops completely. Just reduce them. Stephane: Not about link flapping. Yasahiro Ohara: Clarifications on slide SPF progression. Hannes Gredler: 15-16 year-old algorithms - should we revisit in IGPs? Acee: I think it is a good draft and vendors can keep current their current algorithms as well if required. Jeff: Need to discuss on list. 5) MRT FRR - Chris Bowers (See slides) Stewart Bryant: Clearly a brand new routing protocol algorithm. Needs to be moved back to experimental status. Must be stagged before IETF endorses this. Yasahiro: I presented a Similar algorithm at INFOCOM (see link on rtgwg list) Russ White: Agrees with Stewart. 6) FAR DCN - Bhumip Khasnabish (See slides) Acee Lindem: Draft needs to indicate how protocol works and how it scales better as opposed to just comparing it with a naive OSPF implementation. Pat Thaler: Infiniband does not work the way it is described in the draft. Alvaro: Has this been implemented? Bhumip: Yes - implemented and deployed. (in the slides) 7) DPS - Arun Arumuganainar (See slides) Tony Przygienda - Is there one CE or two CEs? Arun: There can bd a single CE with two separate paths or two CEs Jeff: Please take it to the list. Alia: Some of the requirements seem to coincide requirements for source/dest routing as presented by Fred Baker. Arun: Will contact Fred. Alvaro: Maybe next time we will have solutions to the DPS requirements/use cases. 8) Transport Independent OAM in Multi-Layer Network Entity (TIME) Qin Wu (See slides) Mailing list has been created to discuss this topic: http://www.ietf.org/mail-archive/web/time/current/maillist.html Stewart Bryant: Confused - layers provide isolation and it is no business of upper layer to know what is happening at a lower level. What is the goal here? Qin: Management plane will make sense of information from all layers. Stewart: Common way of reporting? Greg Mirsky: Correlation among layers with applicability to Network Overlay applications are the primary goals. Stewart: Design so OAM will be layers? Greg: This is a pre-BOF discussion to define problem(s). Stewart: I'm even More confused. Dan Romascanu: It is all about information correlation and reduction. Rob Shakir: Worried about alarm suppression. Can we get some standardization across configuration and reporting for OAM mechanisms? Alia Atlas: Potential WGs that have clear protocol pieces are more focused than a BOF for problem definition. Need much more clarity on what protocol enhanced chunks would be covered by the BOF. Pat Thaler: Saw presentation advocating OAM commonility but it was not included (Tissa Senevirathne presentation). Qin: No overlapp between drafts. Greg Mirsky: Tissa's YANG model is not OAM model but just ping and traceroute model. Need to be careful with ITU overlapp for transport mode OAM. Qin: Spoke to Scott Manfield on collaboration with ITU on OAM. Nobo Akiya: Doesn't agree with everything but sees some potential for future work. 9) Wrap Up: Alvaro, Jeff