Abstract:

Current proportion of P2P traffic in the network and the pressure to the network are analyzed in the first, and the problems of network traffic identification technology in practical applications are pointed out. Then,the realization of the principle of deep packet inspection and flow characteristics recognition technology are presented. A P2P traffic identification system model, which combines the deep packet inspection technology and the flow characteristics recognition technology, based on the open source cloud computing platform Hadoop and MapReduce distributed parallel computing architecture, is put forward. The experimental results show that, this model can effectively identify P2P traffic, and possesses a recognition speed faster than the single identification in the face of large flow.

Key words: cloud computing, P2P traffic classification, DPI, TLI, MapReduce