The Research And Implementation On A P2P-CDN Hybrid Content Distribution Structure

With the development of Internet technology, streaming media and other services become an important part of people’s live. Content distribution and network transmission is an effective way for us to obtain the information. From a technical point of view, there are mainly two kinds of technology for content delivery. They are content delivery network and P2P.The main idea of the content distribution network (CDN) is to deploy servers that are close to the user’s location, allowing the user to obtain the required content with the fastest speed and the shortest distance. But it is still a centralized server/client architecture. It uses pre-set fixed server to meet the requirements, but also because these rigid requirements and high costs, it can not be used widely and a lot of network resources are waste. The non-profit sites like schools, government agencies which have a large number of users can not affort the high cost. Peer to peer network(P2P) technology breaks through the traditional C/S mode. Every node is not only the server but also the client. The nodes in the architecture access resources they need from other nodes and transmit the content to other nodes. In this paper, based on the idea of mixed structure of P2P-CDN, we designed a new architecture for content distribution.In this paper, using the idea of mixed structure of P2P-CDN, and fully considering the characteristics of the network topology and network resource access, we designed a mixed level of P2P-based content distribution network structure named LI-CDN (Locality-aware and Interest-focus CDN). The structure takes into account the characteristics of locality-awareness and interest-focus, and the logical structure and network topology mapping. According to the network topology, the nodes in the structure is divided into more than one locality. And according to the different interests of the node, we devide the locality to some sub-localities. This makes content aggregation more closely, and resources can be found and transfer faster. Based on these two characteristics, we designed the entire structure of the underlying network construction, the establishment of the pile nodes, and by providing the functionality of the landmark, the directory node can quickly get its own locality and obtain hierarchical information. We also designed the extension in network coverage layer to enhance network manageability, and the directory node layer node key management program and resources search program.The entire structure is divided into two layers, the directory node layer and the content node layer. In this paper, for the directory node layer, we give a multi-level lookup scheme to enhance the search efficiency, and a detailed description of the contents’information store and update algorithms between the nodes. The content node layer uses the gossip program. We also introduced the preference list to reduce the service pressure of the directory node is also given. Finally, by the simulation experiments, we verifed the performance of our new structure.