DRAFT Minutes of the IP Flow Information eXport (IPFIX) WG 69th IETF, Chicago, Tuesday, 24 July 06 submitted by Nevil Brownlee and Juergen Quittek (co-chairs) based on notes from Dave Plonka The text messaging log is available here: http://www.ietf.org/meetings/ietf-logs/ipfix/2007-07-24.html The meeting agenda and slides are available on the IETF69 'Session Agendas and Presentations' web page. Meeting minutes/notes for IPFIX meeting at IETF69 Tue Jul 24 13:00 CDT 2007 submitted by Dave Plonka -- Juergen reported that we now have six drafts (the original four, Biflows and Reducing Redundancy) in the RFC Editor Queue. IESG comments on Protocol Guidelines is being addressed, Testing has completed WG LC, it will be revised before being submitted to IESG. This meeting was focused on possible new WG items. Criteria for accepting them were summarised as: 1) What is the need, why is this needed, what is its application? 2) Is it based on code, not just an idea (at this point) 3) Do we have committed people to review the drafts/documents (e.g. 3 per each proposal) -- Paul Aitken reviewed the changes in the next version of the Testing draft. Benoit Claise reported on progress with the IPFIX MIB. Not many changes are needed now, next draft expected within 3 or 4 weeks (for last call). Dan Romascanu commented that this and the Testing draft, uses 'copy and paste' sections, rather than referring to one document? He suggested making things more consistent by keeping them in one place in future drafts. -- Brian Trammel l gave a presentation on changes to IPFIX's use of SCTP Streams, and his proposal to remove the restrictions placed on their usage by the Protocol draft. This was discussed at some length; we decided that (after we have discussed the following on the mailing list) we will remove the Protocol Draft from the RFC Editor queue, make the changes to it, rerun the WG and IETF Last Calls, and resubmit it to IESG. That will avoid having an 'update' RFC that implementors and users would need to refer to. No other changes to the drafts will be considered. We will also make corresponding changes to the Implementation Guidelines and Testing drafts, and re-run their Last Calls. There was strong consensus for this at the meeting, -- Benoit Claise presented a proposal for 'IPFIX Export per SCTP Stream' Briefly, each template set should be sent on same stream as associated data. That solves problem about not being able to determine loss per template ID and collecting process' job is now easier because it can assure that templates will arrive before their data records. It could, however, lead to stream exhaustion if many templates are in use. There was support for this as a WG item, with the suggestion that it should become either an Informational of an Experimental RFC. If it becomes widely used, it could later become an option in a revised Protocol RFC. -- Benoit presented the Configuration Data Model for IPFIX and PSAMP draft; it proposes an XML-based configuration data model, a prototype system is running. In the discussion, it was pointed out that other working groups, e.g. netconf, are also working on similar proposals; we should focus on producing English-language descriptions of the information elements (as well as XML, cf the IPFIX Information Model). Once the model is defined, we could consider protocols - perhaps IPFIX could be one of the first WGs to benefit from using a standards-track configuration protocol. No-one present raised any objection to working on such a model. -- Brian Trammel presented the latest version of his IPFIX-Based File Format. This could require the Extended Types Information (see next presentation, below). There was consensus that this should be a WG item. -- Elisa Boschi presented 'Extended Type for IPFIX Enterprise-Specific IEs' This would provide a way to provide data types for enterprise-specific IEs, allowing a Collector to do more than treat them as opaque objects. Discussion centred on how useful this would be. There was little consensus, this will be discussed further on the mailing list. -- Atsushi Kobayashi presented 'IPFIX Mediators,' a hierarchical system for collecting data from many IPFIX probes. People from two large networks commented that they have an operational need for this. However, this is clearly a large topic, not ready to be a WG item just yet. Participation via the mailing list is encouraged. -- Hitoshi Irino reported on some measurements of efficiency gains from changing the ordering of IEs in templates. Consensus was that further work is needed (would like to see measurement results from other sites), this item should continue on the mailing list. -- Tanja Zseby presented Flow Selection Techniques for IPFIX and PSAMP. There is a preliminary implementation, but this proposal is in its early stages. Participation via the mailing list is encouraged. -- Olav Kvittem gave a brief report of some interesting work on QoS Measurement for flows. Audience comment was that one should use existing standard metrics where they existed (e.g. IPPM), and that new IEs can be tried out as Enterprise IEs - if they prove useful they could then become Standard IEs. The IPFIX mailing list is a good place to share ideas on this. -- The Chairs thanked presenters. They will prepare a list of proposed new WG items, for discussion on the mailing list, and possible inclusion in a revised WG charter. The meeting adjourned Tue Jul 24 15:08:43 CDT 2007 =====================================================================