Research On The Network Information Sharing Technology Based On P-Grid

Client/Server architecture is one of the most popular distributed computing schemes in recent years. It divides tasks between computers with different roles, but its centralized infrastructure becomes an obvious limitation in decentralized environments. The resources are stored at a few servers whose amount is very small compared with the total number of computers in the systems. When the systems grow larger, it will definitely become the bottleneck of network bandwidth and system performance. In order to provide continued and reliable access to resources, the servers have to run complicated load-balance and fault-tolerance algorithms locally. Meanwhile, the limited capacity of servers also bounds the amount of sharing information. With the fast development of broadband network in China, users who have great amount of resources will still be kept at the edge of the Internet. At the same time, some P2P information sharing applications like Napster and Gnutella are getting popular throughout the Internet, which made people pay more attention to the P2P technology.In recent years, P2P technology is widely used in file-sharing applications. P2P offers a peer to peer method for information sharing, which has become popular among Internet users. The potential of P2P technology is well noticed as well. But in China, many P2P applications are based on central servers like Napster architecture, few of them has innovations. MMIT lab of East China Normal University began to do researches on a new P2P technology, called P-Grid, and intend to build an info-sharing service platform based on the technology. Consequently, the author made a deep research into the P-Grid technology. The research was focused on the info-search and system build algorithms in P-Grid, which can establish connections between peers strictly through local interactions without any central control. According to the basic idea of P-Grid, peers exchange their information and split the search space when they meet each other. At last, each peer is only responsible for an interval of the whole search space and makes responses to the query requests for the data in this interval. The split of search space can be described as a virtuebinary search tree and the info-search process is similar to the search in the binary tree. As P-Grid provides a directed search method, each time peers forward requests to those peers who are more probable to be responsible for the queries. This greatly enhances search efficiency. Moreover, P-Grid allows data replication among peers, which further improves the robustness of the system. Thus P-Grid can be applied in some environments where network connections are unreliable and possibilities of peers to be online are very low. Based on the research, the author put forward a simplified model of info-sharing system based on P-Grid, which divides the system into four modules: Client, Server, P-Grid and Communication. The author also defined the communication specification in P-Grid system based on XML technology. In the end, the author analyzed the performances of search and exchange algorithm in P-Grid, which proved that it has great predominance and prospects to apply P-Grid technology into info-sharing applications.