You're viewing an archived page. It is no longer being updated.
RIPE 53
RIPE Meeting: |
53 |
Working Group: |
Test Traffic |
Status: |
Final |
Revision Number: |
1 |
Chair Henk Uijterwaal
Scribe Rene Wilhelm
A. Administrative Matters
Henk Uijterwaal chairs the meeting. No changes to the agenda.
Minutes from RIPE 52 were approved.
B. Consumer Broadband Monitoring, A Proof of Concept - Luca Deri
http://www.ripe.net/ripe/meetings/ripe-53/presentations/cbm.pdf
Luca Deri presents the work done at University of Pisa (Italy) on developing a prototype framework for measurements on consumers' broadband connections.
Comments/questions:
Daniel Karrenberg remarks he feels Luca and his students have done good work, it is something he wished we already had two years ago.
Speaking on a personal basis, he remarks this is certainly going in the right direction for a full CBM Service. Daniel asks what happens when the central box goes down. Do current probes store all results until that machine comes back up? Luca answers only the last result is stored, measurements which were not fetched by the server are lost. Daniel suggests enhancing the scheduling mechanism with a mode where the measurement period is not exactly every five minutes, but sees some stochastical variation; e.g. a poisson distribution with an average of five minutes. Selection of the mode could be under control of a parameter.
Niall O'Reilly is also enthusiastic about Luca's work; he would like to have the source code as he uses an old Linux box as a router on his home network; no need for a modified openWRT package. Daniel Karrenberg asks how easy is it to get openWRT to run, what is the success rate? Luca answers that thus far, the success rate is 100%. The recipe varies from router to router. For Linksys devices, uploading new software is just a matter of clicking in the web interface. For Asus, you have to use tftp. Other brands not known (yet).
C. Work on the Multicast Proposal - Franz Schwarzinger
http://www.ripe.net/ripe/meetings/ripe-53/presentations/multicast_monitoring.pdf
Franz Schwarzinger reports on the work he has been doing as an intern at the RIPE NCC on adding support for multicast monitoring to the test-boxes (as proposed by Gert Doering in April 2006)
James Rice comments that the work on multicast is useful to collaborate on and invites Franz to talk to him after the session concludes.
D. TSC Clock Work - Rene Wilhelm
http://www.ripe.net/ripe/meetings/ripe-53/presentations/ttm_tscclock.pdf
Rene Wilhelm showed the first results obtained with the implementation of the TSC clock (Darryl Veitch et.al, University of Melbourne) in a TTM box.
Daniel Karrenberg asks Luca Deri if a TSC clock could also work in the embedded devices of the consumer broadband monitoring project.
Luca replies he will look at it. Some boxes use a MIPS processor, not clear if these have a similar register. Others use a x86, it might work.
E. TTM Future TF, Status and Discussion - Andrei Robachevsky and Henk Uijterwaal (RIPE NCC)
http://www.ripe.net/ripe/meetings/ripe-53/presentations/ttm_future_wg.pdf
Henk Uijterwaal presented the RIPE NCC's ideas on future directions for TTM. The Working Group and the Task Force formed at RIPE52 are asked for feedback.
Comments:
- Niall O'Reilly comments that the proposed new contract where all service fees are payed in advance, so there no recurring costs. This will make it easier for his organisation to become a TTM customer.
- Ruben van Staveren remarks that enforcing a (3 year) hardware lifecycle could generate a conflict with keeping the grid of Test Boxes up
and running. Henk replies that a transition mechanism to the new model would have to be worked out; there is no intention to break the current system of functioning Test Boxes apart. - Simon Leinen comments that the new charging scheme makes sense but he has his doubts on the three year write-off. His organisation has two Test Boxes, these are now four years old but still working fine.
- Henk clarifies the three year period was picked by looking at current practices in industry in writing off computer equipment. He agrees that not all systems are useless after three years; a good balance needs to be found.
Z. A.O.B.
No other business.