Minutes of the IPFIX meeting at IETF 87 in Berlin

About 36 people present, plus 6 on jabber

Scribes: Paul Aitken (on Etherpad at http://tools.ietf.org/wg/ipfix/minutes)
  and Chris Innacio

Meeting chairs: Nevil Brownlee, Juergen Quittek

Juergen opened the meeting, and gave the IPFIX status update:
- IPFIX Protocol, 5101bis
    approved as Internet Standard, 17 Jul 13

Nevil Brownlee presented the Link Layer IEs draft:
- Kobayashi-san has made changes as suggested by Pat Thaler in
  Orlando.  She has reviewed the new version, and reported that
  she will send Kobayashi-san a list detailing a few more needed
  changes.
= After that the draft will be ready to submit.

Brian Trammell presented the IPFIX Mediation Protocol draft:
= -06 version to be published shortly, now ready for WGLC

Paul Aitken presented the MIB Variable Export draft:
- mibFieldOption proposed as alternative to Extended Field Specifiers (EFSF)
- Brian Trammell: this is backwards compatible with current IPFIX
- Juergen Quittek: but it adds an extra level of lookup in collectors
- Benoit Claise: would like to compare EFSF vs option templates
= This needs more discussion on the list

Paul presented "IPFIX Extensions"
- These must solve real problems, i.e. have real use cases
- EFSF would support many use cases that would otherwise need many new
  IEs, e.g. Unobserved Fields, MIB OIDs.
- Johan Kogel: likes it, but it may end up duplication some IEs
- Agreement that using Option Templates to extend IEs is inelegant,
  but we already do it a lot, e.g. RFC 6313 (structured data)
- Paul: it's orthogonal to the 'Option Templates' approach

Abdelkader Lahmadi presented 'Metering Process Location'
- A set of IEs for specifying host location
- Could use EFSF, but need to know times when location changed
- Peng Fan: we should consider whether users will want to install
  IPFIX on their mobile 'phones
= Continue discussion on list

Toerless Eckert presented 'Signaling Flow Characteristics'
- Making a set of IEs that Applications can use to export
  information about their flows
- How many IEs?  ~20 to 30
= Discuss this idea with IE Doctors

Ramki Krishnan presented 'Flow-Aware Packet Sampling'
- The draft has been extensively revised since Orlando
= Its aims could be achieved by registering this as a new
  sampling technique, i.e. value of the flowSelectorAlgorithm
  IE.

There was no Other Business.

The Chairs reminded the WG that
- We need to consider IPFIX Extensions - i.e. EFSF - carefully,
  and reach consensus on a new charter item to work on it
- Any other proposals for new work need to solve a real problem,
  and require something more than changes to the IE Registry

The meeting finished at 1137