Minutes of the IPFIX meeting at IETF 82 About 19 people present Scribes: Chris Innacio & Nevil Brownlee Nevil opened the meeting, pointing out that two drafts from our current charter are still uncompleted: - PSAMP MIB (depends on change to IPFIX MIB, see below) - Flow Selection (many changes since Quebec meeting, needs new WG LC; IPR issues being explored with draft authors and AT&T) We expect that these two will be finished soon. The next part of the meeting discussed items from our new charter. All of these are expected to be republished as new (-00) IPFIX WG drafts, we believe that some of them (as mentioned below) are now ready for WG Last Call. Shingo Kashima sent apologies for the link-layer monitoring draft, so it was not discussed at this meeting. Brian Trammell presented the "IE Doctors" draft; several people commented that the draft explains clearly why some existing Information Elements are 'bad,' it needs some examples of 'good' IEs. The authors will add this into the WG -00 version. Brian presented the "IPFIX Aggregation" draft, WGLC will start when its WG -00 version is published. Benoit Claise presented two drafts that begin the process for advancing the IPFIX base RFCs along the Standards Track. The first was 5101bis, IPFIX Protocol. Many of the proposed changes are 'editorial,' e.g. responding to errata confirmed for 5101. Others are more technical, especially those relating to Template Management and Security for transport protocols other than TCP. In particular, there was considerable discussion of the need to do interoperability testing of DTLS over UDP. That discussion will continue on the IPFIX list, everyone please read this draft carefully and post comments on it! The second was 5102bis, IPFIX Information Model. After discussion, the meeting agreed that we should remove the list of IEs from this draft, i.e. it should simply point to the IANA IE Registry as the definitive document for the IPFIX Information Elements. The XML schema for that registry was also discussed - perhaps a snapshot of it should be in this draft, with the comment that IANA maintain it and may need to change it from time to time. Another issue raised was the 'state' of IEs; states 'current,' 'deprecated' and 'not recommended' were suggested. The state machine for this should probably be in the IE Doctors draft, and it's not clear whether an IE's state should be displayed in the IANA IE Registry. Discussion will continue on the list. Benoit presented the Exporting MIB Variables draft. Discussion centred on the need for a new IPFIX data type for SNMP variables, how best to export variables from SNMP tables, and how best to represent SNMP variables in IPFIX templates. Brian pointed out that we need to be careful that whatever mechanism we adopt needs to allow for future IE namespace growth in a generic way. Discussion will continue on the list. Benoit presented the 5815bis (IPFIX MIB) draft; the only change here is to create a new OID registry for ipfixSelectorFunctions. WGLC will start now (i.e. without waiting for the WG -00 version). The IPFIX Mediation Protocol draft was not presented or discussed, since it's authors believe it is ready for last call now. WGLC will start when its WG -00 version is published. Brian Trammel presented his comments on RFC 6313, Structured Data. It would be useful to add data type information to structured information; e.g., add an options template with the 'type' information, and then extend the structured data types to include another field that points to the options template 'type' information. This discussion will continue on the list. Benoit presented his draft on Exporting Application Information, which has attracted considerable interest on the IPFIX list. Chris Inacio has volunteered to host a 'registry' of Application Identifiers that members of the measurement community could contribute to. Benoit has asked for AD support of this draft as an IETF Individual Submission. There was no time left to discuss Andrew Yourtchenko's Cisco IEs draft. The meeting finished at 1725.