Estimating P2P Traffic Volume at USC

With the rise of peer-to-peer (P2P) file sharing applications there has been an increasing interest in understanding the popularity and use of P2P. In this study, we look at P2P use on the University of Southern California’s campus network throughout a 14-hour period. We quantify the volume of traffic from P2P activity as well as the number of campus hosts involved in P2P at USC. Since port-matching techniques often fail for P2P applications, we estimate traffic based on both port-based and connection-pattern based techniques. We do not have access to packet data and so these measures provide only bounds on P2P traffic. In addition, while we identify P2P sharing, we cannot comment the types of data being shared (either music or data, restricted or freely available). We find that 3–13% of active hosts on campus participate in P2P, and that this traffic accounts for 21–33% of the bytes transferred to and from our campus.