Because P2P sharing of video files uses a lot of bandwidth, some ISPs
have tried blocking or throttling bandwidth for P2P traffic, such as
Australia’s Exetel in 2006 and America’s Comcast in 2008. Each faced
revolts from customers demanding net neutrality. Both providers have
since backed off their restrictions as direct video streaming has
surpassed P2P as the majority user of Internet bandwidth. A possible
future resolution would be for P2P software to schedule background
downloads for the middle of the night and early morning when ISP traffic
For TCP/IP traffic, including Internet video streamed directly from
servers, most ISPs implement HTTP caching. This benefits streaming video
services such as Hulu, Netflix, Vudu, and YouTube. That is large
capacity storage that holds recently accessed data in anticipation of it
being requested by other subscribers. This reduces the amount of
bandwidth that ISPs need from the Internet backbone, in turn keeping
their costs down. P2P traffic uses UDP or TCP/IP ports other than HTTP
and therefore is not included in HTTP caches.
Shortly after their experiment with deprioritizing P2P traffic, Exetel
implemented P2P caching. Thailand’s largest ISP, True Internet, was an
early adopter of P2P caching in 2006 and reported an immediate 60%
decrease in total network traffic. True Internet’s P2P cache vendor is
Israel-based Oversi. With an investment from Cisco, Oversi have built
solutions for ISPs worldwide. Newton, Massachusetts based competitor,
PeerApp, serves another large share of the ISP market.
Cambridge, UK based P2P cache vendor, Velocix/CacheLogic (acquired in
2009 by Alcatel-Lucent), has focused its business more on the content
providers. They have provided content delivery network (CDN) solutions
to BBC, Verizon, and Bollywood.TV among others. PeerApp claims that
return on investment in their products is realized within months.
Fig. 2: Server-streamed video traffic, with TCP/IP caching.
The Pirate Bay—the world’s foremost torrent search engine until it was
charged with assisting copyright infringement and shut down by a Swedish
court in 2009—grew from 10 million to 25 million users over the course
of 2008. uTorrent, the most widely used BitTorrent client software,
estimated 52 million users in 2009.
The Fiber to the Home council reports that the year 2000 was when P2P
file sharing first grew to be the single largest type of Internet
traffic. In 2004 CacheLogic estimated that 62% of consumer ISP traffic
is P2P related. A 2007 study by German traffic management equipment
vendor, Ipoque, estimated that 49-89% of internet traffic is P2P
transfers during the day and about 95% at night.
Traffic management equipment vendor Ellacoya (acquired by Arbor
Networks), estimated in 2007 that 37% of network traffic is P2P file
sharing and 41% is direct HTTP streaming video. YouTube, in particular,
accounted for 10% of network traffic. Traffic management equipment
vendor, Sandvine, estimated that P2P traffic grew from 41% to 44% of
network traffic from 2007 to 2008. Also, in 2008 Arbor networks
estimated that 20% of ISP subscribers used P2P file sharing.
According to the 2009 Cisco Visual Networking Index, in 2008 P2P file
sharing represented approximately 50% of all Internet traffic. The
report predicts a continued rise in P2P file sharing traffic. However,
it estimates that the percentage of Internet traffic for P2P file
sharing will drop to 20% as direct video streaming grows. Direct video
streaming from YouTube, Hulu, BBC iPlayer, and the like is estimated to
have represented 14% of all Internet traffic in 2008. This grew rapidly
to 27% in 2009. The Cisco report predicts that video traffic overall
will represent 91% of all Internet traffic by 2013.
Fig. 3: Historical and projected consumer Internet bandwidth