RIPE 41

RIPE 41 - Test-Traffic WG
Januari 17 2002 - 9:00
Amsterdam, NL
------------------------

Chair: Henk Uijterwaal (RIPE NCC)
Scribe: Mark Santcroos (RIPE NCC)
Attendants: Approximately 30

Agenda:
-------
A) Administrivia
1) Appointment of Scribe
2) Attendance Sheet
3) Agenda Bashing
B) TTM status report/Henk Uijterwaal
C) TTM analysis update/Rene Wilhelm.
D) Bandwidth Measurements/Mark Santcroos.
E) Analysis of End-to-End Delay Measurements/T.U.Delft & Henk Uijterwaal
F) Direction of Group
G) Ripe 42: Call for agenda items
H) AOB


A. Administrivia:
-----------------
Meeting opened by Henk Uijterwaal at 9:05. He replaces Keith who
unfortunately couldn't attend.

The agenda as posted on the mailing list was approved.

The minutes of the previous RIPE meeting, RIPE40 are approved.


B. TTM status report/Henk Uijterwaal
------------------------------------
Slides at http://www.ripe.net/test-traffic/Talks/0201_RIPE41_HU/

Henk reported on the status of the project at the NCC. New services have
been added and more boxes sold. Please refer to the slides for details.

Question from the audience:

Q1: What is the intended use for the 2nd nic?

A1: Sites with more upstream links, don't measure internal network, better
compare the two upstreams. (Also refer to
http://www.ripe.net/ripencc/mem-services/ttm/Talks/0110_RIPE40_HU)


C. TTM Analysis Update/Rene Wilhelm
------------------------------------
Slides at http://www.ripe.net/test-traffic/Talks/0201_RIPE41_RW/

Rene explained about the improvements of the data analysis software. He also
presented 2 case studies which were good examples of how the test-box can be
used to find and solve network problems. No questions were asked.


D. Bandwidth Measurements/Mark Santcroos
----------------------------------------
Slides at http://www.ripe.net/test-traffic/Talks/0201_RIPE41_MS/

Mark presented the new bandwidth measurements currently in beta-test. It is
estimated that this will be a full service around RIPE42. Questions from the
audience:

Q: Did you try this on satellite links?
A: Not really, we just tried this between our test-boxes, there might be
satellite links there but we didn't look for them.

Q: Do you see a difference satellite/earth link?
A: Don't know, will look into it.

Q: Measurements are Layer 2 independent, what does that mean?
A: Because it is just normal traffic it is not likely to be
re-prioritized.

Q: All boxes, scheduling.
A: Will do as many measurements as possible in 24h. Probably tune this,
or make it configurable.

Q: What you are measuring is physical capacity of the bottleneck.
A: Yes, it is protocol independent.

Q: Protocol dependent depends on stack but might be more interesting.
A: Yes, but no standard for measuring them yet.


E. Analysis of End-to-End Delay Measurements/T.U.Delft & Henk Uijterwaal
------------------------------------------------------------------------
Slides at http://www.ripe.net/test-traffic/Talks/0201_RIPE41_TUD

Henk presented a study done together with the TU Delft on the TTM data. He
showed that the minimum delay between hosts can be explained by sum of the
delay in each hop and the propagation speed of the signal through the cables.


F. Direction of Group:
----------------------
Item was skipped because Keith was not here.


G. Ripe 42: Call for agenda items:
----------------------------------
At RIPE40, the topic of netflow came up and it was suggested that a
presentation from an expert would be useful. Keith has meanwhile found a
speaker for this, unfortunately he couldn't attend RIPE41. It was suggested
to postpone this to RIPE42 (and RIPE43).

Henk asked if people were interested in a re-run of the SIGCOMM tutorial on
"how to use measurements in daily ISP's operations", as an EOF session. There
was some interest in this, Henk will take this up with the RIPE chair.


E. AOB:
-------
No other business.