Internet Society Frontpage

Events Membership
About the Internet Standards
Publications  Public Policy
About ISOC Education

About the Internet 

Become an ISOC Member

Internet Evolution

Question 2: Decline in P2P Traffic Volume

As the data indicate, streaming video is now the largest category of Internet traffic. Peer-to-peer (P2P) traffic is still increasing but at a slower rate. This question is concerned with whether a potentially superior technology for background bulk data transfer (P2P) is being overtaken by less-efficient tools for content delivery as a result of nontechnical concerns. Is alignment with existing business models trumping technical efficiency in the marketplace for high-bandwidth content delivery?

In considering this question, the superiority of P2P technology for certain types of applications and file transfers is clear to some.

Is the relative decline in P2P traffic volume indicative of the triumph of business models over technology?

Geoff Huston: The use of P2P models by the open-source community as their preferred method of distribution as compared to the archaic and woefully inefficient models used by the mainstream vendors for software upgrades is a clear illustration of the effectiveness of P2P to perform the heavy lifting when widespread nonreal-time content distribution is the goal.

The relative decline in usage is indicative of the effectiveness of the threat of overwhelming penalties applied on an almost random basis. The copyright holding industry is staring down the barrel of irrelevance and rather than adjust their business model to new technologies and utilize distribution channels that are overwhelmingly more efficient and cheaper than their existing hard, media-biased distribution systems, they are consistently adopting a position of fortressing up anachronistic business models of bygone years and using lobby pressure against the political system to enact Draconian penalties for anyone who would dare to transgress their copyright assets.

Bill St. Arnaud: Yes. P2P decline is largely attributable to the Gestapo-like tactics of the record and movie industry to thwart piracy.

The film and music industries have certainly lobbied hard to restrict and reduce incidences of illegal file sharing over the Internet. However, the demonstrated shift in traffic patterns may also indicate that P2P tools are reaching a natural peak in their user base. As more convenient tools become available, users, regardless of the underlying mechanisms, will tend to adopt them.

Kenjiro Cho: I do not see it as business models–versus-technology.

It shows a success of video content over the Internet; initially, P2P was used among technically savvy users but it is shifting away to Web-based services to reach much broader users.

The data are also undoubtedly a reflection of the impact of ISP traffic-management practices, which seek to restrict the use of P2P applications during certain periods or for certain classes of subscriber. The extent to which these practices are the root cause of the observed data is an interesting question.

Alissa Cooper: It is likely that a combination of factors contributed to the findings: the growth of streaming video and direct download, network operator traffic management practices that target P2P for throttling, P2P protocol innovation, and encryption of P2P application traffic.

These last two factors, and the wealth of continued innovation in P2P applications and protocols, would seem to indicate that technology continues to play an important role in the development of P2P. Recent developments in the IETF likewise signal continued interest in P2P innovation, most notably with the formation of the ALTO, LEDBAT, and DECADE working groups in the time since the study. Because P2P technologies have proven to be immensely effective in transferring large volumes of data and localizing data transport, they are likely to remain a fruitful area of software development for many years to come.

The decline in the P2P share of traffic volume does hold an important lesson for the future of traffic management. Because the mix of applications on the network will continue to change, traffic-management policies that target particular applications will continue to require upgrades while potentially becoming less effective as the targeted applications decline in popularity. This constant game of cat-and-mouse with applications developers and its diminishing returns may compel network operators to pursue traffic-management solutions based on generic traffic characteristics rather than application or protocol signatures.

Of course, as we discussed earlier, there is plenty of application-layer innovation going on, and that includes P2P applications. Given the pace of change on the Internet, maybe this ‘shift’ in application preferences is a short-lived phenomenon and not enough to tell us anything about the overall direction of travel for Internet evolution.

Joe Touch: A single report isn’t a trend, and a single trend isn’t a triumph. If they were, we would be talking about the triumph of Archie over FTP. Triumphs in the Internet are short-lived.

There are many possible reasons for the data reported, not the least of which is the use of negotiated ports to avoid firewall filtering (e.g., to circumvent copyright enforcement). Another viable reason is that some of what P2P used to do is now being serviced by YouTube, iTunes, and NetFlix, to name a few.

Jon Crowcroft: The range of relationships between service providers and customers is richer than ever; P2P is just one (network-centric) model.

Jonathan Zittrain: I think it’s hard to draw firm conclusions from raw traffic volumes. P2P changes could indicate a new generation of protocols that favour transmission over subnets, which would result in less traffic at network interconnection points. And a minute of video requires much more traffic than a minute of reading, so it’s hard to say the Web is taking a back seat to a non-Web P2P protocol on that basis alone. The right metric is mindshare, and meaningful opportunity for someone with a message to convey it using various modalities.

Have Your Say

How would you answer this question? Provide your answer below.

< Previous QuestionNext Question >

Post a comment