P2p Distributed Mass Storage System For The Download Experience

While the information in the Internet is exploding, the network storage technology attracts more and more concerns and recognitions. However, the primary problem of this technology is how to make good use of leisure personal computers which were once ignored to form a large virtual storage space, which can share oceanic resources to the clients. With the progressing of Peer-to-Peer technology, the network storage technology captures a good opportunity to make this ideal come ture. Constructing a large scale distributed mass storage system, which has high availability, high reliability, and great fault-tolerance, is one of the research highlights in recent years.This thesis researched and concluded P2P and distributed storage's history and status quo. Then it designed and implemented a P2P Distributed Ocean Store System(PDOSS) based on users's download experience by these technologies.The major work of this thesis is shown below:First, the PDOSS's architecture and model were designed and implemented. The details of file operations( publish, download, update and delete) were also designed and implemented. PDOSS takes advantages of both of structured network and unstructured network which makes it have the characteristics of rapid locating and good flexibility. PDOSS is of a two-layer network architecture. The first layer is a structured overlay network formed by server group. The second layer -- client layer, which is constituted by client nodes, is a BitTorrent network which makes the clients can share their resources with each other conveniently. The advantage of this architecture is making managing and controling clients easier because of the existence of these servers. Meanwhile, the structured network will avoid many defects such as the single point failure.Second, the route protocol was designed and implemented after P2P network's structured route algorithm– Chord was studied and researched.Third, a dynamic multi-domain replica management strategy was presented based on the two technologies mentioned before. The concept domain is based on the different network speed between LAN and WAN. The system gathers and calculats the clients's download experiences. And then it dynamically adjusts the replicas's redundancy and location. This strategy improves the clients's experience and decreases the expenses of the servers.Fourth, various kinds of timer were designed and implemented to improve the system's reliability, which would ensure that the system can find out the broken down servers and recover the data as soon as possible.