Research And Implementation Of Distributed Storage Strategy Based On MooseFS

Currently, the rapid development of the Internet makes information resources present a trend of explosive growth. Network and data has penetrated into every aspect of people’s life. Cloud computing and the application of the Internet produces huge amounts of data, which puts forward higher requirements for the storage and analysis of data. The problems existing in the traditional storage mode also gradually appear, such as poor manageability of data, bad scalability of the system, limited storage capacity and high cost for upgrade and maintenance.Distributed storage technology can effectively solve the above problem. The key technology is the application of the distributed file system. The distributed file system technology can meet the need of storage capacity for large-scale data. It also has the advantages of redundant storage, high transmission rate and can be upgraded online. All of the characteristics can make the system meet the demand of storage and the access of a large number of users at the same time.The thesis studies on the distributed storage technology. The working principle of MooseFS is analyzed in detail. The thesis puts forward the improved scheme for the storage strategy of the distributed file system to optimize the performance of system. For the scenario of the video on demand application, a distributed storage system for video on demand based on B/S architecture is deployed on MooseFS platform. The system can share the pressure for file storage and user access on the server.First of all, the thesis studies on the typical distributed file system such as GFS and HDFS by analyzing their structure and working principle. MooseFS distributed file system is selected for further study. On the basis of the study on MooseFS working principle, detailed analysis is carried on for the original storage strategy of MooseFS. The source code is parsed. Aiming at the problems existing in the original storage strategy, an improved storage strategy is innovatively presented using the server performance evaluation function. When the system carries on the new storage strategy, it will introduce more parameters to evaluate the server’s performance, so as to balance the storage load between the server clusters and optimize the storage performance of the system.Secondly, a distributed storage system for video on demand based on B/S architecture is deployed on MooseFS platform. The design of the system is divided into two parts. One part is the module design for video on demand based on B/S structure. The other part is the design of the distributed storage system on the server. The design of the module for video on demand adopts B/S architecture. Users can realize the video playback function through the browser instead of some complex application software, which could save resources for the system. For the deployment of distributed storage system, a server is mounted to the MooseFS system as the client node in order to realize the distributed storage of the video files.Faced with the growing demand for data storage under the situation of rapid development of cloud computing, the thesis conducts the research on technologies related to the distributed storage based on MooseFS. In order to improve the resource utilization of the system, the storage strategy in MooseFS is improved, which balances the data distribution between the storage nodes. A distributed storage system for video on demand is implemented to meet the need on storage capacity. It can also effectively share the pressure between the servers for storage and access to the system. The study of the storage strategy in MooseFS and the implementation of the distributed storage system in this thesis mean a lot to the research on distributed storage technology and the performance optimization of server clusters.