Agenda Bashing- no adds. Introduction ------------------------------------ Chris Morrow and Peter Schoenmaker are new WG chairs. Thanks to DMM and GH for previous work. Updated Milestones: ------------------------------------ 1) Define standard for BGP data collection, methodologies, and format. Geoff: Draft exists for MRT dump data around somewhere. Geoff mentioned two formats for data storage. Major router vendor's ASCII dump, and MRT format - binary dump. Multi-prot-BGP and 4-byte ASN support missing from current draft. Cisco and MRT have different update formats. Cisco uses syslog MRT diff... MRT only has one reference implementation by RIPE. Chris: commented that this milestone is about more than data format, also about data gathering methodologies. Chris mentioned that the different methods for configuring/data gathering lead to different appearing results. Rutiger: asked about what else is considered - whether further analysis of data is planned, including performance monitoring etc. Chris said that this is a possible further step for this item. Peter mentioned that GROW is focused on global routing tables rather than ASN specific interior tables. 2) Document causes for global internet route table growth Peter: Leverage the standard gathering techniques to accurately track growth 3) Document causes for global internet route stability/instability Peter: Build on Geoff's good work and analysis. Geoff: Instability is often local artifacts that also have local impact. What he thinks are important are the differences between how the network is trying to react to failures, with how the routing system (protocol) tries to adjust. Geoff suggests documenting some taxonomies around environment changes and how that impacts protocol changes. Lixia: 3rd bullet is a direction/marching order of where to go. She says Geoff's point is consistent with the 3rd bullet. Geoff: my point is a refinement/addition to the 3rd bullet. Example is how route-flap dampening didn't solve the correct problem. Was a solution created in the protocol with out complete understanding of the root causes that turned out to be a mistake. 4) Document common practices for interprovider connectivity Work Spaces ---------------------- Clear scope of work limited to working with operational techniques/technologies, not protocol design. A few community proposals, interconnection best common practices Danny: Interdomain should be emphasized since it will break first. Suggests GROW should provide feedback to IDR on examples of good/bad technologies (implementation feedback for optimization) that have been deployed. Peering Best Practices ----------------------- Joe Abley presenting: Engineering and Operations about peering. Not about policy/religion about techniques. Including BGP policies, Undesired behavior, peering protection BGP resilience, etc. Outline finished, text is being written, 00 draft planned in next month or so Question is whether GROW is a good place for this work. Ted: What is exact purpose of this document, seems like a lot of work.... Joe: Really a list of things to avoid/consider to keep from making common mistakes. Better name is Interconnection between ASNs. Peter: Document seems interesting, GROW looks forward to seeing the entire document. Open Discussion ---------------------------- Additional comments: Danny: Look at some other Groups that have work languishing, and as an example communities work is important. Multi-path and community attribute stuff is interesting. Geoff: IDR's community string work has been based on suspicions/ideas not fact... Danny: Optimizing BGP convergence time is a real operational issue that needs to be more understood. Create a feedback loop between operational views to the protocol design. Geoff: Data points are really hard to gather, only now are we understanding the impact of protocol changes to BGP on the network.