Research And Application Of Distributed File System Fastdfs

With the development of the Internet, information on the Internet are growing exponentially, especially the information in the form of picture, video, audio. It’s more difficulty to store and manage the information. Traditional storage architecture can not meet our requirements, after investigation, using distributed file storage system can solve the problem.After analyzing and comparing the distributed file system at home and abroad, this paper select FastDFS as the underlying storage system. FastDFS can provide file Storage, access and management. This paper improves the banlancing algorithm by considering storage space utilization, connections and performance instead of one single factor-total storage space. Tests are conducted in three cases:one is the case only considering storage utilization, the second one considering storage utilization and connections, and the third one is the case considering all the three factors including storage utilization, connections and performance. And the degree of the load imbanlance of three cases are0.0010,0.0096,0.01147respectively. Improved load banlancing algorithm works better than the original algorithm in load banlancing. Although the first case works best, it is more near to the realistic scenarios after adding connections and performance factors. Then this paper makes optimization for the new algorithm. The primary factor changes from storage utilization to performance. Although this leads to poor load imbanlance, it improves performance.The second work in this paper aims at appling FastDFS to the image processing system, which makes it more effective especially for mass small files processing and settles down the fail-over problem. Also it increases the accessing speed. This paper introduces the image processing system in three aspects from architecture design to the key technology’s achieve. After the testing by. simulating the actual environment, the system can satisfy the needs of receiving capability, concurrent processing capability and real-time responsiveness, high scalability requirements for the image processing system.In this paper, the prototype system comes from the DongBei KaiYa company’s actual business needs. This paper improves the load balancing algorithm of FastDFS and applys it to image processing system. The results show that the performance of the system improves a lot.