RIPE 74 MAT Working Group Minutes

Date: 10 May 2017, 14:00-15:30
Chair: Nina Bargisen
Scribe: Suzanne Taylor
Status: DRAFT

A. Introduction: Welcome, Scribe, Jabber, Agenda

Nina welcomed everyone to the session and thanked the scribe, chat monitor and stenographer. She mentioned that the RIPE 73 draft minutes were sent on the mailing list and are available on the www.ripe.net website.

She asked whether there were any objections to the draft minutes from the RIPE 73 MAT Working Group session. There were no objections and she declared the minutes approved.

She asked whether there were any changes or additions to the agenda; there were none.

B. Fast.com

Sergey Ferorov

This presentation is available online:
https://ripe74.ripe.net/presentations/105-Ripe74-Fast.com-1.pdf

Thomas Schmid, DNF-Verein, asked whether Sergey had used Quick for their measurements. He said that unfortunately they cannot reliably use Quick on the browser and that, on the server side, they don't have it implemented. He said they are familiar with the protocol and might want to play with it in the future.

An audience member asked whether Sergey planned to include the option to test for IPv4 or IPv6. Sergey said that they hadn't included it yet because 90% of users don't know what IPv4 or IPv6 is, but that they were working on advanced features and planned to include these options.

Geoff Huston, APNIC, asked what TCP flow algorithm Sergey was using on his server. Sergey said that they may run multiple versions at any given point, including BBR and their own implementation of congestion control.

Geoff responded that you can see different things depending on which algorithm you use, and so asked whether Sergey was trying to report in his ability to reach a user with a particular congestion control algorithm, or whether he was trying to report on what the network is able to sustain as a flow control. Sergey said that they work TCP maximisation in order to maximise throughput and so they use multiple implementations, but that Geoff's question is something the team debates. He said that currently the number is more a representation of how much you can achieve over the network, versus how much the network can sustain if you use any possible trick.

Geoff said that with BBR, that's not true – that BBR is incredible aggressive about taking what it can do without queing and knocking everything else out of the way. He said this is why he's interested in knowing what Sergey is trying to measure – what you can achieve if you're “nice to your neighbours” or what you can achieve if you “kick them in the mouth”.

Sergey said that in general you shouldn't run anything else when running a speed test, so ideally the two numbers would be the same. He said they're not trying to measure the network when there is competing traffic.

There were no further questions.

C. v6 Health Metric with NOMA

Leslie Daigle

This presentation is available online:
https://ripe74.ripe.net/presentations/98-20170508-NOMA-at-RIPE74-FINAL.pdf

Shane Kerr, Oracle, asked whether the intention is for researchers or operators to be able to access the data. Leslie said that the overall intention is to publish the metrics, but not all of the raw data.

Shane said that DNS-OARC has been collecting and contributing data for many years and that it is very useful for operators and researchers, and that thinks Leslie's efforts were really good and he looked forward to seeing where she ends up with it.

Bela Bödecs, BIX, said he had a technical suggestion to use a logarithmic ratio rather than a pure value for the ratio to make comparison easier. Leslie said she would take the suggestion under consideration.

There were no further questions.

D. RIPE Atlas News

Robert Kisteleki

This presentation is available online:
https://ripe74.ripe.net/presentations/74-ripe74-mat-atlas-update_v2.pdf

Ondrej Caletka, CESNET, asked whether there was a replacement for the RIPE Atlas anchor hardware planned since things don't look good for the Soekris box. Robert said that this was a surprise even for the supplier and that the good news was that the supplier has an inventory of 50-60 boxes, so the RIPE NCC should have a supply for six months or so. He said the RIPE Atlas team is looking into a next generation solution. Ondrej said that CESNET had something that might be suitable and Robert said that the team was aware of that and would like to speak to Ondrej about it.

Brian Nisbet, HEAnet, said he would try to rehabilitate his v1 probe, and asked how a user would opt in to the wireless measurements. Robert said the feature had not yet been rolled out on the UI because it was still being tested, but that it would be soon. Brian said he would be very interested to opt in as soon as the feature is available.

Randy Bush, Internet Initiative Japan, asked Robert to find a good next generation anchor hardware solution, as the cost of installation was significantly more than double the cost of the hardware itself.

There were no further questions.

E. TraceMON: Network Debugging Made Easy

Massimo Candela

This presentation is available online:
https://ripe74.ripe.net/presentations/84-TraceMON_RIPE74.key

Christian Kaufmann, Akamai, said he likes the tool a lot. He asked how the system chooses the probes. Massimo said that, currently, this is tightly coupled to RIPE Atlas, so it uses the same probes the user chose when creating their measurement, but that in the future, he would like the tool to intelligently choose a subset of these probes to use.

Christian then asked whether there is a version on TraceMON on the RIPE Atlas website, rather than just GitHub. Massimo said that the code is available on the RIPE Atlas website and that it can also be used as a widget if the user doesn't want to download the code themselves.

Daniel Karrenberg, RIPE NCC, said that it might be a good idea to limit the number of hops from the target that are shown, because sometimes he's really interested in how close it is to the target, rather than trying to visualise 1,000 targets.

Massimo said that this feature is implemented, but that he's received feedback that users want to know the source of the traceroute, so he must choose which features to implement based on all users' needs. For now, this is what he's chosen to implement for now in the beta tool. Daniel said he's not interested in knowing the source and asked Massimo to make it possible to simply see the last three hops, for example. Massimo reiterated that the tool is just in the beta phase and will continue to be developed.

Alexander Azimov, Qrator Labs, said he really liked the tool but that it was currently not working on the RIPE Atlas website. Massimo said he would look into the problem.

There were no further questions.

F. Russian Hackathon Examples

Alexander Isavnin

This presentation is available online:
https://ripe74.ripe.net/presentations/97-ISAVNIN-RussianInternetMeasurementsHackathons2017-04-25.pdf

Sergey Myasoedov, Internet Citizen, asked how likely it was that a governmental structure would use the hackathon results to build another digital border for the Internet. Alexander replied that the goal was the opposite – to better understand how to oppose legislation to protect critical and non-critical infrastructure.

Sergey asked whether Alexander was communicating this goal when promoting the hackathon. Alexander said the first hackathon was organised with the Internet Protection Society, which opposes the current government. He said that Sergey should know that he doesn't support the current government's Internet regulation.

There were no further questions.

Z. AOB

Nina reminded everyone that there were PC elections and that voting would end at 17:30 tomorrow, so encouraged everyone to vote.

She also reminded everyone that Friday's plenary begins at 9:30 with a session on diversity in the RIPE community and encouraged everyone to attend.

There were no other items and she closed the session.