A Method Based On Campus Environment File Server Cluster System Design And Implementation

As the network technology develops, the informatizated application system of information has been stepping into modern lives. More and more people tend to use computer and network technology to improve their efficiencies, which becomes an indispensable part of daily office work. File service is a popular network service by Internet. File server, as a kind of basic large-capacity storage media, plays a significant role in enterprises and campuses, and has a higher requirement in perfonnance because of its massive throughout in unit time. For years, many NSPs(Network Services Provider) endeavor to improve the perfonnance of large-capacity or extra-capacity file server, to meet the requirement to the higher-performance file services.By making good use of server resources with different configuration and architecture in campus, from the point of researching users of the file server, this thesis centers on how to improve the comprehensive usage and file-service perfonnance in file server, which is organized by clustered architecture. The primary task is to design a clustered file server system based on campus network, to implement the high-performance transmission of large-capacity files. Algorithms, including overload-balancing algorithm and file-scheduling algorithm based on prediction mechanism, through which the balanced allocation for file-upload requests and file scheduling between server nodes are implemented, are designed to improve the overall performance. Otherwise, this system also implements the unified management for storage resources, with the consequence of providing a higher transmission rate to file upload and file download. By a long-time testing the file server system, which was built in the laboratory, the result shows that file transmission and tasks scheduling work fine in this system with files allocated and intelligently scheduled between nodes.