RIPE 76 MAT Working Group Minutes

Thursday, 17 May 2018, 16:00 - 17:30
WG co-Chairs: Nina Bargisen; Brian Trammell
Scribe: Alun Davies
Status: Final

A. Introduction

Video available at:
https://ripe76.ripe.net/archives/video/135/

Brian Trammell (MAT WG Chair) opened the WG session and welcomed participants. After some preliminary procedural comments, Nina Bargisen (MAT WG Chair) and Brian then welcomed the first speaker.

B. Opportunities and Challenges of Ad-based Measurements from the Edge of the Network [15 min]

Patricia Callejo

Video available at:
https://ripe76.ripe.net/archives/video/137/

Brian Nisbet asked how the technology Patricia had presented relates to the growing use of ad-blockers. He asked what she sees for the future of ad-blockers. Patricia responded that these issues will always be there – she doesn’t see them going away. Another audience member asked about measurement targets changing when people put more money into the ads. Patricia responded that she didn’t think so and that it’s more of a case of which websites host the ads. Brian asked how the methodology the speaker had described can be used – e.g. to run his own measurement campaign. Patricia said she would gladly show him all the measurements they’d collected.

C. Real Time Monitoring and Troubleshooting of Web Browsing Sessions [15min]

Antoine Saverimoutou

Video available at:
https://ripe76.ripe.net/archives/video/140/

Biran asked Antoine if the tool he’d discussed was on GitHub yet. Antoine replied that anyone could mail him and he would send the tool to them. Brian asked if this was to reduce ovrheads. The speaker answered that the tool would be needed to measure across all browsers and that they’re still trying to identify relevant parameters for measurements made with the tool during browsing sessions.

D. Multi-Gbps HTTP Traffic Analysis in Commodity Hardware Based on Local Knowledge of TCP Streams [15 min]

Carlos Vega   

Video available at:
https://ripe76.ripe.net/archives/video/141/

A speakers from the audience asked why the speakers measurements avoid the TCP connection, which would help performance. The speaker explained that they rely on trust anchors to deal with the huge amounts of data involved. The speaker from the audience pushed the question of whether Carlos and his team are taking into account user agents. Carlos responded that there are various factors to take into account here and that the reasons for doing it their way ultimately boiled down to fitting what they’d developed into a bigger process of network management.

E. Perceptual Evaluation of Web-Browsing [15 min]

Diego Neves da Hora

Video available at:
https://ripe76.ripe.net/archives/video/143/

One member of the audience commented that the speaker should be cautious of certain issues that can arise from using PLT measurements. He then asked how the speaker named the objects that were being visualized. Diego acknowledged the issues around PLT and pointed out AJAX JavaScript would not work with the approach he’d presented because of those issues. As for naming, he mentioned that they’d used the HTML image on the resource timing to get a match. The member of the audience raised a few further questions, and Diego made clear that they were looking at things very much from a browser point of view. 

Brian Trammell asked how much variation the team had been getting from the JavaScript widget they’d been using. The speaker and Brian had a back and forth on the issue and it was agreed that such effects would only have a very small impact on the measurements.

F. RIPE NCC Tools [15 min]

Robert Kisteleki

Video available at:
https://ripe76.ripe.net/archives/video/144/

Robert was asked a number of questions about the prospects of VM anchors and software probes. He discussed some of the issues around developing this kind of approach with RIPE Atlas and assured the audience that there would more progress to report on this in the near future. One speaker asked whether VM solutions were planned to replace hardware counterparts, Robert said that this would be unlikely and that VMs were meant to complement the existing hardware anchors.

G. Measuring Routing Insecurity at the Global Scale [10 mins]

Andrei Robachevski   

Video available at:
https://ripe76.ripe.net/archives/video/147/

Daniel Karrenberg asked how the presenter had defined resolution. He pointed out some issues that could arise from different ways of doing this and advised the speaer against certain approaches. Andrei agreed with Daniel’s points.

Randy Bush suggested that the speaker should also be looking into active measurements. The speakers asked for clarification and Randy provided further details.

Brian Trammell asked a question aimed at putting the work back in Working Group. He asked whether there’s a working document for the approach presented and whether that was available on the mailing list. Andrei replied that this had not been done yet as they’re still collecting feedback, but that such a document would be shared with the working group when ready.

Brian thanked the speaker.

Z. AOB [5 min]

[There was no other business to be discussed.]