Minutes IETF97: ippm
minutes-97-ippm-00

IPPM Agenda - IETF 97 - Seoul, Korea
====================================

Monday 14 November 2016 - 15:50 - 17:50 KST (UTC+9)

Main Agenda
-----------

All agenda items available at
https://datatracker.ietf.org/meeting/97/materials.html under "ippm" Meeting
chaired by Brian Trammell Notes taken by Barbara Stark. Jabber scribing by
Jason Weil.

Time  | Speaker        | Item
------+----------------+-----------------------------------------------------
15:50 | Chairs         | Note well, intro, status, agenda bash
16:10 | K. Nieminen    | Invited Talk: Network Neutrality Problem Statement
16:25 | A. Morton      | draft-ietf-ippm-[initial,metric]-registry
16:50 | G. Fioccola    | draft-ietf-ippm-alt-mark
17:00 | K. Pentikousis | draft-ietf-ippm-twamp-yang (skipped)
17:15 | G. Mirsky      | draft-mirsky-ippm-twamp-refl-registered-port

### WG draft status

https://www.ietf.org/proceedings/97/slides/slides-97-ippm-0-chair-slides-00.pdf

draft-ietf-ippm-6man-pdm-option
- Handling secdir review comments: can DO be used to expose information to
on-path elements? how about DO outside ESP? Discussion between Brian and Nalini
Elkins. secdir review

- draft-ietf-ippm-2330-ipv6:
no update, waiting on 6man.

- draft-ietf-ippm-model-based-metrics-08:
Submitted to the IESG for publication
Al Morton is working on this.

- draft-ietf-ippm-twamp-time-format-00:
waiting on write-up (Bill)
Just a few things to do and Bill will write things up.

### Invited Talk: Network Neutrality Problem Statement, K. Niemenen

https://www.ietf.org/proceedings/97/slides/slides-97-ippm-1-net-neutrality-measurements-regulatory-use-case-and-problem-statement-00.pdf

Klaus Nieminen presented slides. Said he would like comments from IETF.

Henning Schulzrinne: There are many types of help that could be provided. What
exactly are you looking for and how much money do regulators have to pay for
building infrastructure to measure ISP networks? Klaus: Hoping to do crowd
sourcing. Henning: Some of this can be very complex. Klaus: Acknowledged
complexity. Henning: In US also noted a lot of geographic distribution. Klaus:
Geography is important but also need to ensure ISP is supplying guaranteed
speeds.

Al Morton: On Slide 3, do you really mean *all* traffic or just all user
traffic? Do you really want network management traffic treated equal with user
traffic? Al Morton: On slide 4, have you published a methodology for speed
metrics? Klaus: ISPs neeed to take various factors into account.

Vinayak Hegde: Problem with zero-rating? How do you take care of that?
Klaus: ISP must be able to deliver promised speeds.
Vinayak: Definitions of network neutrality vary from EU to US to India, etc. Is
this a problem? Klaus: No. We do not need a common definition. Brian/Vinayak:
But this causes problems with metrics definition if everyone isn't defining
network neutrality the same way.

Barbara Stark: Do you intend to run measurements for all users in order to
confirm guarantees? Klaus: No

Henning: Most providers don't lie because it's not a good idea. Most net
neutrality violations are because there is incentive. You don't have to look
too hard and make measurements too expensive. In many cases you can rely on
fact that when users have good tools, they will report the problem. So the
tools can be available for users to use when they have a problem.

Jason Weil: Last slide, clarification on monitoring aspect. Are you really
saying you don't want ISPs to monitor traffic? Or you don't want them to act on
knowledge obtained by monitoring?

J. Ignacio Alvarez-Hamelin: Have these tools already been developed or are you
in process of developing them? There are many issues with these tools and how
are you handling them? Klaus: Development has started.

### draft-ietf-ippm-[initial,metric]-registry, A. Morton

https://www.ietf.org/proceedings/97/slides/slides-97-ippm-2-draft-ietf-ippm-initialmetric-registry-00.pdf

Al Morton Presented slides.

On Slide 6, suggests new namespace registry for metrics.
Brian (as contributor): I think that's not enough. There are other metrics than
performance metrics. Al: So we would be specifying metric and subspace (instead
of having a separate namespace).

Questions?
Brian (as chair): Is this what's needed for LMAP?
Jason Weil: I think this is being written for LMAP. As long as info model stays
in sync we'll be fine. We need to have consensus call. Spencer Dawkins: How
does this relate to model based metrics? Al: We did presentation on that.
Brian: Rephrasing Spencer's question to clarify. Al/Brian/Spencer: Conclusion
that all is well. No need to worry. Al: Need feedback from people implementing.
Or at least imagining themselves as implementers.

### draft-ietf-ippm-alt-mark, G. Fioccola

https://www.ietf.org/proceedings/97/slides/slides-97-ippm-3-alternate-marking-method-for-passive-performance-monitoring-draft-ietf-ippm-alt-mark-02-00.pdf

Giuseppe Fioccola presented the slides.

Stewart Bryant: Want to focus on what measurements we really need to do.
Giuseppe: Agreed
Proceeded to end.
Suggested WGLC.
Brian asked if anyone objected to starting WGLC. No-one objected.
Al: You had average, mean delay, and that's the least useful, because you
really want to know the extent. Stewart: Need mean and variation. You can get
what you're looking for by also providing variation. Brian: Take that to the
list, so I can make WGLC happen after you resolve.

### draft-mirsky-ippm-twamp-refl-registered-port, G. Mirsky

https://www.ietf.org/proceedings/97/slides/slides-97-ippm-5-udp-port-allocation-for-the-receiver-port-in-twamp-draft-mirsky-ippm-twamp-refl-registered-port-01-00.pdf

Greg Mirsky presented the slides.

At end asked for adoption.
Brian said not adopting today. Asked for hands of who had read. Few hands. We
need more people to read draft before we adopt.

Lightning Talk Queue
--------------------

### draft-bhaprasud-ippm-pm, S. Jacob

https://www.ietf.org/proceedings/97/slides/slides-97-ippm-6-packet-loss-measurement-model-draft-bhaprasud-ippm-pm-00-00.pdf

Sudhin Jacob presented slides.

Greg Mirsky: You need to have consecutive measurements around a threshold.
Service degradation definition is different per service. Sudhin: Will add.

### draft-mizrahi-ippm-multiplexed-alternate-marking, T. Mizrahi

https://www.ietf.org/proceedings/97/slides/slides-97-ippm-7-passive-monitoring-using-a-multiplexed-marking-field-draft-mizrahi-ippm-multiplexed-alternate-marking-00-00.pdf

Tal Mizrahi presented slides.

Stewart Bryant: I assume you're going to do long delay measurements and short
samples. Tal: Yes. Stewart: Wants to make sure hardware is not touched. Brian:
Remind me where this goes? Tal: Proposed to use DHCP Header.