DEAPFI: A distributed extensible architecture for P2P flows identification

Abstract

Identification of various P2P flows accurately is fundamental for network management. To obtain P2P flow information in real-time and low price, a distributed extensible architecture for P2P flows identification called DEAPFI is proposed. DEAPFI decentralizes functions of flow identification and statistic on different PC, which is composed of Identification System (IS) and Statistic System (SS) respectively. IS integrates many independent P2P flow identification techniques and arranges their running order following optimizing sequence. By utilizing the identification results of IS, SS can extract P2P flow information from NetFlow records generated by NetFlow enabled equipment, which effectively reduces the requirement of PC's operation ability. The experiment results of the prototype system verified DEAPFI can reduce the disposal burden of IS and SS, and provide a feasible scheme for online identifying P2P flow information in high speed network.