Bittorrent manipulation in selected countries

 

The data pattern for Canada is interesting because it shows that BitTorrent throttling by Bell Canada actually increases substantially in late 2010 and 2011, after a 2009 CRTC decision regulating Internet Traffic Management Practices. While Cogeco seems to abandon the practice in 2010, Bell Canada’s use of throttling of its retail service increases, despite a notice to its wholesale customers that its wholesale service “may not be subject” to such throttling after November 2011. High levels of BitTorrent throttling by Rogers Cable also remained unchanged by the 2009 CRTC regulations. Bell Canada and Rogers were both among the world’s top 10 throttlers in 2011 and 2012. Canaca, TekSavvy and Primus also show increases in 2011. Rogers has promised that 50% of its customers would be free of DPI-based throttling by July 2012 and that they will end it entirely by the end of the year. As of the 1st of March 2012 there is no sign of any change.

Canada

Glasnost data, Q1 2009 - Q1 2012

On these interactive charts, each line represents an ISP. The vertical axis shows the percentage of times Glasnost tests indicated that the ISP was manipulating BitTorrent using DPI during a 3-month period. The horizontal axis is time, running from the first quarter of 2009 to the first quarter of 2012. Mouse over any line 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 bars or as bubbles.

Germany

Glasnost data, Q1 2009 - Q1 2012

Kabel Deutschland stands out as the only major user of application-specific throttling in Germany. All other Internet service providers do not seem to use it. For Kabel, there is a dip in the positive rate in the third and fourth quarters of 2011 but this is probably due to measurement errors caused by server problems in the MLab platform.

Great Britain

Glasnost data, Q1 2009 - Q1 2012

Hong Kong

Glasnost data, Q1 2009 - Q1 2012

Taiwan

Glasnost data, Q1 2009 - Q1 2012

Brazil

Glasnost data, Q1 2009 - Q1 2012

The UK is one of the few countries where BitTorrent manipulation appears to be on the rise. DPI measurements for the BT Group in particular increased progressively throughout the 3-year period. Talk Talk’s use of it seems to have declined, but it is still consistently above the threshold indicating some form of manipulation. Test results for O2 UK move erratically slightly above and below the error threshold, making any conclusion difficult. Virgin Media, on the other hand, seems to have altered its policy and increased DPI-based intervention in the 4th quarter of 2010. Measurements for Demon Internet are scarce, but seem to be completely free of DPI-based application discrimination since Q3 2009. The academic network JANET (not shown on the chart) aggressively blocks BitTorrent.

Hong Kong was a uniquely hostile environment for P2P protocols in 2009. All three of its broadband providers, HKBN (City Telecom), Hutchison and PCCW, tested positive for very high levels of BitTorrent manipulation that year. Despite boasting of a capacious fiber network, HKBN was one of the world’s top throttlers from 2009 to 2011. But PCCW seems to have eased off in 2010, and HKBN may also be doing so in 2012 (the small number of tests in the first quarter makes it difficult to be sure). Hutchison fell out of our measurements altogether in 2010, 2011 and 2012, as too few of its users conducted tests. In general, the number of tests for Hong Kong has dwindled substantially since 2009.

Taiwan, like Hong Kong, shows extensive use of DPI-based manipulation of P2P protocols. Taiwan has one or two of the world’s top throttlers every year, including its nationwide academic network, which blocks BitTorrent. TFN consistently tests positive for BitTorrent manipulation, and appears in the top 10 list along with Taiwan Academic Network. Smaller Taiwanese providers, such as Union Broadband and Yeongjualeh Cable TV, showed 50% - 70% rates of manipulation in the periods for which we have sufficient data. HiNet does not seem to be engaging in DPI-based manipulation of BitTorrent, but there are only sufficient measurements for parts of 2009 and 2010. Sony Network, alone among Taiwan’s providers, seems to have reduced or eliminated the practice in late 2009.

Brazilian ISPs have a mixed pattern, with overall a relatively high level of DPI use but with some indications it is declining. Oi (Portugese for “Hi!), the largest telephone company in Brazil, showed a consistent pattern of throttling moderately all through 2009 to 2011. Embratel, Net Servicos, Intelig Telecomunica and Empresa de Infovias also show high rates of manipulation in 2009 and 2010. Oi has a markedly lower rate in the first quarter of 2012 which, if continued, indicates a policy shift. Global Village Telecom may also be reducing the extent of its intervention. Telefonica Brazil, on the other hand, does not seem to have ever used DPI for BitTorrent manipulation, with most quarterly measurements at or below the error threshold.

Poland

Glasnost data, Q1 2009 - Q1 2012

Poland, like several other Eastern European countries, has experienced a definite shift away from DPI-based throttling. UPC Poland, a cable infrastructure-based network, was one of the world’s top BitTorrent manipulators, but it seems to have changed course some time in late 2011. UPC affiliates in many other European countries have also turned away from DPI-based throttling. In  Poland, TP Group also seems to have moved from moderate levels of throttling in 2009 and 2010 to negligible levels in 2011 and 2012. A similar pattern is shown by cable provider Vectra.

United States

Glasnost data, Q1 2009 - Q1 2012

BitTorrent throttling by most US ISPs ceased after an FCC ruling in August 2008 that declared Comcast’s actions to be against its 2005 Internet Policy Statement. Only wireless ISPs Clearwire and Hughes showed high levels of BitTorrent manipulation after August 2008, but Clearwire substantially reduced it in 2009. But the Comcast Order was reversed by the courts in April 2010. This data shows that no US ISPs have increased their use of DPI-based throttling since April 2010, despite the absence of any network neutrality regulation in the US. We lack sufficient data about Hughes in 2010 and 2011. After the second half of 2010, the only US ISP showing high levels of DPI was Cogent, but the number of tests for Cogent is very small, just over the minimum for inclusion in the table.