Identification And Analysis Of P2P Traffic Of Private Business Based On Behavior Characteristics

With the development in last several yearas, the applications adopting P2P technology have gradually played an important role that could not be ignored, in the process of Internet changing the way of daily life. However, for Internet Service Providers, the huge consumption of bandwidth resource has affected the overall quality of service. Although Internet Service Providers have work out several research and solutions on identification and management of P2P traffic, there is room to deal with the problem better. Especially with the rapid development of big data, it is practical to make calculation on large-scale data set, the combination of these new technologies will bring out many breakthroughs in research of traditional fields. Facing huge P2P network traffic, we can take advantage of the cloud computing technology to breakthrough the bottleneck of computing and observe the distribution of P2P traffic in a more complete and macro perspective, which is significant for summarizing features and proposing traffic identification programes in huge amounts of data. In this thesis, we firstly introduce the widely used distributed computing platform Hadoop with it basic structure and related components. Secondly, taking a specific P2P application as example, we introduce the communication process of P2P and present a traffic identification program based on MapReduce Mdodel on the angle of behavior characteristics of roles. This method can effectively deal with massive amounts of network traffic data and adapt to the rapid increasing scale of data and processing requirements. Then, we present the detailes design of distributed P2P traffic identification and analysis system based on Hadoop, focusing on some important components such as relevance model, concurrent merge model and role recognition model. Finally, some characteristics of P2P traffic are presented and analyzed combined with identification results.