IPPM-WG @ IETF 76
=================

Chair:   Henk Uijterwaal
Scribes: Erik Burger
         Matt Zekauskas (remote participant)

1. Administrativia.
   The Chair opened the meeting with the usual stuff.  The co-chair
   could not make the meeting and sent his apologies.  Erik Burger
   voluntered as scribe.  The agenda was approved without changes,
   the speakers for items 4 through 8 will all present through WebEx.

2. Status of drafts and milestones
   The chair gave an overview of drafts not discussed today:

   * draft-ietf-ippm-more-twamp-03.txt: published as RFC5618
   * draft-ietf-ippm-multimetrics-13.txt: RFC5644
   * draft-ietf-ippm-reporting-04.txt: this draft passed WGLC,
     it is now on the chair's desk and will be passed to the IESG
     shortly.
   * draft-ietf-ippm-framework-compagg-08.txt: Ready for WGLC right
     after this meeting.

3. Reporting drafts. (Al Morton)
   a. draft-ietf-ippm-reporting-04.txt: see previous item.
   b. draft-ietf-ippm-reporting-metrics-00
      This draft became a WG item, comments received on the individual
      submission version have been included in the -00 version.  The
      author is looking for more readers.

4. Metrics composition drafts status. (Al Morton)
   * draft-ietf-ippm-framework-compagg
   * draft-ietf-ippm-spatial-composition

   Looking for WGLC on framework draft, Henk says this will be done soon.

   Spatial composition updated based on what agreed to.  Resolved a comment
   on an early section, which was the case of computing e2e performance
   for multiple segments of a path even if e2e routing is not working.
   This case is now mentioned.

   Ruediger Geib said he had some comments that he will post to the list.

5. TWAMP features
   a. draft-ietf-ippm-twamp-reflect-octets-03 (Morton, Ciavattone)
      Al briefly outlined the proposed changes that the draft makes
      to the protocol.  It is essentially ready to go to WGLC but the
      readership has been small.   As this makes significant changes
      to the protocol, a review would be good.  Unfortunately, the
      folks active in the initial version have all moved to other
      things.  The chairs will discuss this further.
   b. draft-ietf-ippm-twamp-session-cntrl-02  (Morton, Chiba)
      This draft is ready to go to WGLC.

6. Advancing Metrics along the standards path (Geib, 20')
   http://www.ietf.org/id/draft-geib-ippm-metrictest-01.txt

   Ruediger presented this draft.  Comments from the floor:

   Matt Mathis: How aggregated was path measurement taken from?  It is on the
   backbone of a major provider, thus heavily aggregated.  MM said we need a
   statistician to understand what happens when temporal resolution taken to
   low values.

   Focus was on delay, just a little on packet loss and jitter. Al: ideally
   think about test plans for validation while developing the metrics on
   throughput.

   Is there interest in the work group for this draft? Al, Reza, Rudi, Gerard
   have been working on the draft.  At the SF meeting, 4 others expressed
   interest but they haven't commented.  Henk will chase them.

   Ruediger is interested in doing live testing, but then sees issue with
   shipping equipment around the world.  There are some possibilities though,
   to be investigated.

7. TCP Throughput Testing (Barry Constantine, 20')
   http://www.ietf.org/id/draft-constantine-ippm-tcp-throughput-tm-00.txt

   This is potential new work, presented to see if there is interest.
   Barry gave an overview of the draft.

   Comments: Reudiger: It is OK to standardize methods and metrics, but
   probably not pass/fail metrics. He does not like idea of putting
   monitoring system on that constantly loads network.  Barry: Intention is for
   this to be internal, turn-up testing. Not targeted for constant monitoring.

   Take to list for interest as a WG item.

8. Burst Loss Metrics/Dick Duffield (20')
   http://tools.ietf.org/html/draft-duffield-ippm-burst-loss-metrics-01.txt

   This draft was presented by Nick.  Comments.

   Matt Mathis: Has not read draft; however, it sounds like the assumption is
   the narrowest link is the link likely to be lossy. That may not be true.

   A comment on Matt's comment: in the draft there's no assumption about
   where loss occurs. I guess Matt has in mind probing with back-to-back packet
   pairs (or trains) and measuring the stretch or compression in order to
   estimate bandwidth.

   The draft methodology is different; the separation between the packets in
   a pair is the temporal resolution at which different loss episodes are to
   be distinguished.

   Nick will do another draft. Submit to WG for consideration.

9. AOB
   No other business came up.