File-aware P2P traffic classification

As the enormous growth in the number of P2P applications in recent years, ISPs are striving to balance between (1) providing the basic networking service for P2P users, and (2) properly managing network bandwidth usage. In terms of the first aspect, it requires ISPs to provide enough bandwidth for a P2P user to get each file in order to satisfy basic needs for all users. In terms of the second aspect, it means that the aggregated bandwidth consumption should also be managed by ISPs. However, current P2P traffic management strategies are unable to satisfy both requirements. Choosing file as the basic management granularity is an effective way to adjust P2P bandwidth consumption because file, as the elementary transfer object in P2P networks, correlates closely with both user demands and network bandwidth. According to the idea, a file-aware P2P traffic classification method is presented in this paper. It steps into the content level which leads to an finer-grained classification than existing methods, and can identify a file and the associated flows. Based on the file-level concurrent flow information, ISPs can adopt more efficient and flexible strategies to manage P2P traffic. We offered two alternatives: limiting the aggregated bandwidth consumption (directly) or the number of concurrent flows (indirectly) that a peer can use to download a particular file. Finally, a real-life trace is measured using our method. Results are shown from the perspectives of peers and files, and proper parameters of traffic management strategies are given.