Test-Traffic Working Group

Minutes Test Traffic BOF during RIPE 31

Minutes Test Traffic BOF
during RIPE 31
Edinburgh
24 Sept 1998
Chair: Henk Uijterwaal
Scribe: Daniel Karrenberg



Agenda
0.  Administrativa.
    - Agenda bashing.
    - Scribe.
1.  Status of the project (HU).
    - Installation of test-boxes.
    - Problems with installation of the antenna and how to solve them.
    - Analysis of the data 
    - Discussion
2.  Discussion on how to proceed with the project in 1999.
    - Are the results that we've produced so-far useful? 
    - What other numbers/results are people interested in?  
    - More test-boxes? Where to install them?  Funding?
    - Other experiments that can be done with the test-boxes.
    - Interaction between the hosts of the test-boxes and the Test-Traffic
      group at the NCC (see also point #3).
      
3.  Future of the Test-Traffic BoF.
     So far, the test-traffic project was in a development stage.  This has
     changed over the last few months when we started to do regular measurements
     of the delays between the test-boxes.  During the development stage, there
     was little interaction between the TT group and the host of the test-boxes
     or the group of people who will eventually benefit from hosting a box.  
     In order to improve the interaction between the TT-group and the hosts of
     the boxes, we suggest to establish a working group to discuss the
     test-traffic project.  In the future, the group can also discuss results
     of other, related, projects.
     In order to establish a WG, we need a document describing the goals of
     the group and a chairperson.  If anybody wants to volunteer, please contact
     me before the meeting.
     In the meeting, we can discuss this idea as well as the exact goals
     of the group.
4.  A.O.B.
-----------------------------------------------------------------------
1.  Status of the project (HU).
Presentation at: http://www.ripe.net/test-traffic
Henk described the current status of the project including the problems
experienced with long antenna cables.  Higher gain active antennas were
tested with little improvement.  The best soloution for these is to use a
low-loss cable type.  We have good experiences with 'Aircell 7'.  There
currently are 12 test boxes fully operational and another 7 in transit or
being set up.  Henk showed a number of plots as well as routing vector
histories and discussed ideas about interpretation of them.  Finally Henk
discussed plans for the immediate future. It was re-iterated that the
current prototype results are available at a password protected web page.
For practical reasons currently all participants have access to all data
as there was no objection to this. If any participant has problems with
this, they should contact the project team. Daniel asked for anyone with
knowledge of or experience with software to store and retrieve large
amounts of time series data to contact the project team. Erik-Jan Bos
suggested looking into SAS. 
2.  Discussion on how to proceed with the project in 1999.
Presentation at: http://www.ripe.net/test-traffic
It was suggested to provide a plot of forward and reverse paths together.
Also plots with visual correlation of measurements and clock
synchronisation quality were suggested.  Daniel asked whether there was
interest for SNMP monitoring and traps of paramterised values. Several
people were interested. 
3. Continuing as a WG
There was consensus that this was useful.  Martin McCarthy of MIDS was
volunteered and appointed as chair by acclamation.