The Research And Implementation Of Server Cluster Construction Based On DFS

We now live in an era of information explosion, the development of information technology, particularly the birth of Internet, generated and processed data has reached a PB (1PB = 1024TB), EB (1EB = 1024PB) or even more, we call the level of data massive amount of data, the question followed is how these mass data storage and how to ensure high availability. Although these problems have been resolved in the past years, but when the data reaches the level of mass, the previous methods can not cope with this change, we must study new ways to solve these problems. After the data reaches to mass level, these problems can only be resolved by using a distributed file system to build server clusters, using a distributed server cluster technology to address the massive data arising from the series of problems.This article first mainstream distributed file systems research, after the analysis of the mainstream of the distributed file system architecture, advantages, disadvantages and applicability, we select a distributed file system to build our Distributed server cluster. In this paper, we studied how to build a distributed server clustering technology based on this distributed file system, using the Linux operating system and Xen virtual machine software to build the entire cluster. This paper also studied the distributed file system installation and configuration technology, operate the distributed file system normal. In the study based on the work to achieve the successful deployment of the server cluster.In a distributed file system, file storage allocation load balancing algorithm is a very important part, in order to optimize the system, improve system performance, this paper selected a distributed file system load balancing algorithm, analyzes the original algorithm deficiencies, for mass data storage needs of the practical application of proposed solutions to improve the algorithm, the improved algorithm has been tested validation requirements to achieve the improvement, the system performance is optimized. This paper obtained the following results:1. In this paper, the mainstream of the distributed file system, each distributed file system architecture, application environment, the advantages and disadvantages and so do a thorough research, select the appropriate system to build this experimental environment.2. Study the working principle of the distributed file system, Distributed File System mastered the various working mechanisms.3. Propose a distributed file system based data storage solutions and build a Xen virtual machine based on Linux and software experimental platform for the future and lay the foundation for further research.4. Studied the file allocation MooseFS distributed system load balancing algorithm, which found insufficient, and its improved algorithm is proposed. After comparing the experimental data, the improved algorithm to achieve the desired results.