Minutes for IPPM at IETF-93
minutes-93-ippm-1

IPPM meeting, 1130 Friday 24 July 15, Athens/Barcelona

0 - Note Well / Agenda

1 - Registries for O/TWAMP (draft-morton-ippm-owamp-registry-01)

OWAMP - registry not properly defined, Al to fix that.

Al at mic, Registries for OWAMP
problem is that bit positions are used now, need new registry
to define these.

ACTION: WG Last Call to start next week.

2 - Active and Passive Metrics and Methods (draft-ietf-ippm-active-passive)

Al at mic. Hybrid = active + passive.
Brian Trammell: Note this definition of hybrid differs from one discussed on
list.
  Observation points at various points in the stream -
  Al: Spatial metrics refers to multiple observation points.
  Brian: Or, passive observe when someone else generates the stream and you’re
  passively observing it Al: That’s an attack! Surveillance is an attack! Brian
  Trammel: But we need a clear definition of 'hybrid.'

Kostas: Observing some other stream is simply passive
  Al: Asks for further review.  List discussion offered a really interesting
  taxonomy.

ACTION: Next: Chair: Need at least one more revision, take to list.

3 - IPv6 PDM Destination Option (draft-ietf-ippm-6man-pdm-option-00)

Nalini: NETMAP/VALE - kernel module for hi-speed pkt I/O.

Al: Source fragmentation?  MTU 1500, but we add another 20 B,
  -> fragmentation.

Greg Mirsky: Consider that as overhead in path MTU Discovery

Brian Trammel: diversity of devices used for testing?  Nalini: None so far.
   Brian: Middlebox work: how useful are IPv6 options?

Philip: Two reports on propagation of IPv6 options, about 10% of options don't
work.
   This is recorded in a v6ops draft.

Unknown: 16 bits field size is too small (wraps too quickly)
  Nalini: We had a long discussion. It was made 16 bits in the interest of not
  chewing up a lot of bytes. Destination and multicast address?   Nalini: Not
  tested yet. Need to see what happens with multicast. Multiple options to
  single destination?  Nalini: Needs testing

Nalini in response to question: Implementation using IPsec must put in-clear
material ahead of any encrypted content.

Nalini: next: Windows version.  Maybe also IBM mainframe, cell phone.

Marius Georgescu Using transition techniques?  Will just get dropped.
  Nalini will add a note about this.  Brian Trammel: Don't get into too much
  detail.

4 - Model Based Metrics (MBM) draft-ietf-ippm-model-based-metrics-06 (Matt
Mathis)

  Draft reworked to make it easier to understand.

  Mirjia Kühlewind: Why different layers in the MBM picture?
  Matt:  Shows different test streams running at same time (?),
    failure of any one will affect the others.

  Engineering tests must pass for measurements to work, but they
  won't be checked as part of normal testing.

  Mirjia Kühlewind: Do you have an implementation?
  Matt: We do have an implementation that was done by an intern. There is a
  reference in the back.  It is open source. Mirjia: Maybe changing the title
  would help? Matt: I can do that.

  Kostas Pentikousis(?): TCP Measurements RFC (RFC 6349) may need updating.
  Al: It wasn't a standard.

  ACTION: Chair: One more rev, then WGLC (estimated around September 2015)

5 - TWAMP Data Model (draft-cmzrjp-ippm-twamp-yang-01)

Who's read the draft?  3 to 4.
Need more people to read it!

Craig: You include SID from RFC 6038?  Need to think about that.

Unknown: List of times (for tests)?  Chair: IP-based security considered
harmful? Kostas: Definitely. Al: We could add that, not too much effort needed.

Way forward: Chair: Re-chartering won't be considered today. We can have this
discussion before Yokohama.  Developments like this do seem to fit the
charter.We won't consider milestones now.

ACTION: Needs new revision meantime.

6 - TWAMP Light YANG data model (draft-mirsky-ippm-twamp-light-yang-00)

Greg Minsky: Looking for others to co-operate on list.

7 - Support of IEEE-1588 time stamp format in TWAMP
(draft-mirsky-ippm-time-format-01)

Greg Minsky: Follow-on from discussion in Dallas.

Richard Foote (Octel-Lucent): TWAMP-light often refers to its appendix, could
we simplify it now?

Greg: Yes, probably should do that. I would be interested to start working on
it. Al: TWAMP-light errata was rejected by Martin Stiemerling, ”go back and
rework the document"

Al: The reason this is in an appendix is because TWAMP light has no security,
but that's mandatory.
  Chair: It needs a proper Security Considerations section
Spencer Dawkins: Is this hopeless to fix?
Brian Trammel: From my understanding, of the stuff that’s implemented, I think
it is hopeless to fix;  if we’re adding in back-end features that if it looks
like TWAMP, it might be possible to say these are some minor changes to the way
that TWAMP works. There is an issues in that there are two divergent ways;
there are two protocols that kind of look like the same protocol. And there is
a divergence on them.

Matt Mathis: The document could have language that says it's worth documenting,
   that it's another TWAMP implementation, but it's dangerous to use
   because of its lack of security.
Brian: That would be an informational document.
Chair: We need to take this to the list.
Al: Just use TWAMP!
Chair: If people who have implemented TWAMP-Light are willing to write an
informational document up to fix ambiguities, I think we would be willing to
consider that as a working group.

8 - A packet based method for passive performance monitoring
(draft-tempia-ippm-p3m-00)

  4 people have read.
  Alessandro Capello, Alternate Marking story ..
    wants to "generalise the alternate marking principle"
    to MPLS, BIER, etc

  Chair: Comments on the list please.

9 - TWAMP Extension to monitor Services KPIs
(draft-sp-ippm-monitor-services-kpi-00)

  One important slide (!)

  Chair: Thanks for the initial brief presentation.
    This looks interesting, consider it as a
    thing to do - start with one or two further cycles.

Fin at 1322.