Based Network Distributed Storage Technology Research And Realization

The durative enlargement of network size, the rapid expansion of network resources, and the increasing requirements of the data access speed and data reliability, make the distributed storage technology to face enormous challenges. Fortunately, the peer-to-peer (P2P) technology provides some fresh light on resolving the problem of distributed storage. Different from the traditional distributed storage technology, the P2P technology does not focus on server devices and storage equipments, but resolves the problem through the full use of user nodes in the network. Therefore, the distributed storage systems based on peer-to-peer (P2P) network have larger capacity, better expansibility and reliability.A P2P based Distributed Storage System is presented in this dissertation, which has better extensibility, security and robustness. The main contributions of this dissertation are as follows:(1) A P2P based Distributed Storage System with two layers architecture is designed. The scalability of the system is enhanced by constructing a two-layer network. The system is composed by content distribution nodes and user nodes. Content distribution nodes store the information of network topology and the indexes of resources, and user nodes store pieces of resources, some redundancies of the network topology information and some indexes of resources. The upper layer is an unstructured P2P network composed by content distribution nodes. Therefore the system can still have good performance in the case that content distribute nodes join and quit frequently. Each content distribution node manages a group of user node. All user nodes compose the under network. This dissertation also describes the detail of the rules for creating and maintaining a network.(2) The Lookahead with Catch (LHC) algorithm is proposed. First, some unstructured P2P algorithms are described and analyzed. Then the LHC algorithm is proposed based on the own characteristics of this system. By means of the simulation, we compared the system with other unstructured P2P searching algorithms, and verified that the LHC algorithm has higher efficiency on resource searching. Therefore the LHC algorithm is used to realize the resources searching module of the system.(3) The information redundancy module is designed and realized. In this system, the content distribution node sends its node information and resources indexing information to its neighbors and the user node it managed. The redundancies are used to avoid the situation that a user can't get the resource, which is caused by the off-line of a content distribution node. Thus information redundancies insure the reliability of the system. As the nodes in the system only share limited storing space, when the content distribution node do not has enough space to store, it will store the information in the use nodes it ruled. A node choosing scheme for redundancy information, which is based on the available storage space of user nodes, is proposed in this dissertation.