Research And Implementation Of Traffic Locality In P2P-CDN Streaming Media

With the development of Internet, there are more and more and more streaming video, which have the properties of high data volume, high-bandwidth, high-traffic and high quality of service. However, the feature of "best effort" for Internet at this stage makes it difficult to realize the applications of large-scale streaming media under the existing network architecture. Meanwhile, the P2P and CDN technology, which may use the technology of peer entry, advance content delivery and cache server, have resolved this problem to a certain extent. But they both has its own problems and weak points, nowadays integration of P2P and CDN technology has been put forward and has aroused wide concern. Through which, it can realize their complementary advantages in enhancing the system performance greatly.As P2P-CDN research one of the key technologies in agriculture, traffic locality, is also commonly known as flow optimization technique, routing technology, is through the application layer control and choice, optimize the system of data distribution efficiency, reduce cross-realm network flow. Therefore this paper research significance lies in the routing strategies to promote P2P-CDN system performance and efficiency.In this paper, by analyzing and studying various existing P2P-CDN system and traffic locality models, the author summarized advantages and Characteristics of them, and then presents a new traffic' optimization model based on biased neighbor selection:peers'popularity modeling. The core idea of this model is to introduce a new kind "popular degrees" assessment system, through collecting the information network to assess performance, and in peers selection when carry on certain sequencing and optimization, promote data distribution efficiency.In this paper, the author finished the implementation of neighbor select by improve the tracker server in BT system finally, and then realized the simple simulation and testing with Python scripts. The testing results show that this model realized the basic nodes sequencing and selection function, but specific model still needs further improvement and ascend.