Network transparency cuts both ways. It can be exploited to engage in surveillance of Internet service providers as well as Internet users. In order to better understand DPI use and the scope of its deployment, the project makes use of crowdsourced network monitoring data. So far, we have used data from a test known as Glasnost, which was developed by German researchers to detect blocking or throttling of BitTorrent and other peer to peer (P2P) file sharing protocols. The detailed workings of the Glasnost test are described in Dischinger, Marcon, et al (2010).

Thanks to an initiative known as the Measurement Lab (MLab), supported by Google, the New America Foundation and the PlanetLab Consortium, the Glasnost test allows end users all over the world to test whether their ISP is blocking or throttling BitTorrent and other protocols. The results are stored and made available to researchers. The Glasnost data begins in April 2008 and continues to the present time. An Internet user who runs the Glasnost test can see whether BitTorrent is completely blocked, slowed down (throttled) or running normally. The Glasnost data was processed using a method developed by van Eeten,Bauer, Asghari, & Tabatabaie, (2010).

This feature of the web site will be developed more extensively as we process more data and develop additional tools for searching and visualizing the results. For now, we show results for the period April 2008 - May 2010.  NEW: You can now download the Glasnost data shown in the interactive table here.

BitTorrent Throttling by US ISPs

Glasnost data, Q2 2008 - Q2 2010

Interactive Table:

BitTorrent Throttling by ISPs, Worldwide

Glasnost data, Q2 2008 - Q2 2010

Each bar represents an American ISP. The left axis shows the percentage of times Glasnost tests indicated that the ISP was manipulating BitTorrent using DPI. Click the “play” button to see how the percentage changed over the two year period. Mouse over any bar to see the name of the ISP and the specific percentage as a label, or click the box next to the name of an ISP on the right to highlight the results for specific ISP(s). Use tabs at the top right of the chart to visualize the data as lines or as bubbles. In this case, BitTorrent throttling by all cable ISPs is virtually eliminated after an FCC ruling in Q3 2008 that declared Comcast’s BitTorrent throttling to be against its 2005 Internet Policy Statement, even though that ruling was later reversed by the courts.

You can view data by country, operator name, time period (quarter), and by the range of valid tests conducted. The column on the far right shows the percentage of times Glasnost tests indicated that the ISP was manipulating BitTorrent using DPI. Glasnost seems to generate false positives of around 10% prior to August 2009 and of 4-5% after that - so some ISPs who do not throttle BT at all may show some positive results. The number of valid tests is important because the more valid tests done, the more reliable the results in the last column. E.g., ISPs for whom we have only 11-30 tests per quarter (only 1-2 tests per week) will be highly variable and thus less reliable than ISPs for whom we have >450 tests per quarter. We do not show results for ISPs with less than 10 results per quarter. Charts visualizing the results for a few selected countries (US, NL and BR) are shown below.

BitTorrent Throttling by Netherlands ISPs

Glasnost data, Q2 2008 - Q2 2010

Each bar represents a Dutch fixed-line ISP. The left axis shows the percentage of times Glasnost tests indicated that the ISP manipulated BitTorrent using DPI. Click the “play” button to see how the percentage changed over the two year period. Mouse over any bar to see the name of the ISP and the specific percentage as a label. Click the box next to the name of an ISP on the right to isolate the result to that ISP(s). Use tabs at the top right of the chart to visualize the data as lines or as bubbles. In this case, cable-modem ISP UPC’s throttling in 2009 generated complaints and was scaled back. Measurements for Zeelandnet are unreliable because of the small number of tests for that small localized ISP.

BitTorrent Throttling by Brazilian ISPs

Glasnost data, Q2 2008 - Q2 2010

Each bar represents a Brazilian (fixed-line) ISP. The left axis shows the percentage of times Glasnost tests indicated that the ISP manipulated BitTorrent using DPI. Click the “play” button to see how the percentage changed over the two year period. Mouse over any bar to see the name of the ISP and the specific percentage as a label. Click the box next to the name of an ISP on the right to isolate the result to that ISP(s). Use tabs at the top right of the chart to visualize the data as lines or as bubbles. In this case, Canadian regulatory proceedings from May 2008 to October 2009 had no impact on P2P throttling levels, even though Canada’s Telecommunications Act gives the CRTC, unlike the US FCC, clear and unambiguous authority over ISPs’ network management practices.