Evaluation of an Efficient Measurement Concept for P2P Multiprotocol Traffic Analysis

The ability to accurately detect peer-to-peer (P2P) network traffic associated with different applications has become crucial for a broad range of network management tasks including application-specific traffic engineering and capacity planning. Since traditional methods like port-based identification have become highly inaccurate, as P2P applications nowadays tend to camouflage their communications, a new measurement concept based on application layer signatures has been developed. In this paper we present the results of a detailed laboratory reference evaluation of our implementation, which validates the detection accuracy of this approach. We further present the results from a P2P traffic analysis at a large campus network, which show that this P2P traffic measurement concept provides the performance and scalability required for practical applications