IPFIX WG meeting at IETF 78, 1300 on 29 July 2010
Minutes by Nevil Brownlee, with thanks to scribe Chris Inacio

IPFIX WG status:
Nevil Brownlee reported that errata 2052-2059 should say 'unsigned32,'
rather than 'unsigned64;' he will liase with RFC-Editor and Dan
Romascanu to get that corrected.  Two IPFIX RFCs (Mediators Problem
Statement and SCTP per-stream) have been approved by IESG and are in
th4e RFC EDitor Queue.

Current WG Documents:
Gerhard Muenz presented the Configureation Data Model, and made it
clear that their have been significant changes.  In particular, the
handling of Selection Processes (SPs) has been simplified.  A new
version (-07) of the draft will be published soon, then we will run
another WG Last Call.

Brian Trammell spoke briefly to the Anonomysation draft; this is waiting
for Nevil to do it's Shepherd writeup.

Haruhiko Nishida reported on the Mediators Framework draft.  The -06
version was published after an editing session at the Anaheim IETF,
-07 and -08 incorporate further improvements.  The draft now has
WG consensus; Juergen will do it's Shepherd writeup.

Salvatore D'Antonio presented the Flow Selection Techniques draft,
making it clear that the draft addresses only flow-state-dependent
sampling.  Benoit Claise pointed out that flow selection will need to
be considered in a 'Mediation Protocol' draft (specifying a protocol
to configure IPFIX Mediators).  The draft editors will publish the
next revision soon, then it will be ready for WG Last Call.

Benoit presented the Export of Structured Data draft, including
comments from four WG reviewers.  After comsiderable discussion, the
meeting agreed that List semantics need to well-defined and orthogonal
to the underlying data structures.  After its next revision, the draft
will be ready for WG Last Call.

Benoit led a discuussion (no slides) of version -01 of the PSAMP MIB.
Help from the MIB Doctors is needed to model the MIB's unsigned64 and
float64 objects (Dan asked for this to be discussed offline, and
suggested we should avoid creating a new Textual Convention).  Apart
from that, the draft needs work on its Security Considerations, and a
set of normative references to the other IPFIX RFCs.  When the next
version is published we will start its WG Last Call.

New (non-WG) Drafts:
Brian Trammell presented 'Exporting Aggregated Flow
Data using IPFIX.' This is very much a preliminary version, pointing out
the issues raised by aggregated flows; it only applies to temporal
aggregation, not to spatial aggregation.  There was considerable
discussion, and comment about the interplay between aggregation and
anonymisation.

Brian went on to present 'A Lightweight Textual Format for IPFIX.'
This is intended as a non-XML represetation of IPFIX Information
Elements for use in simple IPFIX Applications.  Benoit suggested it
might be useful for "IPFIX Doctors" (cf MIB Doctors), or for people
who don't need to know IPFIX's IANA details.  This was an initial
presentation, if you are interested please send comments to the
IPFIX list.

Shingo Kashima presented 'Information Elements for Link Layer
Measurement.'  Dan suggested that existing FrameType Textual Conventions
could be suitable for describing a link type, and commented that
this should use a 16-bit value.  Paul Aitken suggested that the
draft authors should check the IPFIX IEs 1-127, so that this draft
could maintain compatibility with NetFlow v9.  AGain, comments on
the IPFIX list are welcome.

The meeting reviewed the IPFIX Milestones, Nevil will ask the
Secretariat to change them.

-----------------