Research And Design For Cloud Storage Systems Oriented To Terminal Applications

With the development of information technology industry and its gradually maturation, its various applications have penetrated into all walks of life in the social, and also make the social productivity level greatly improved. More important, it has brought unprecedented convenience and benefits for people’s life, work, and study. Behind this, information technology brings the explosive growth of data, and there is badly need for cheap, fast and safe way of data storage. In recent years the rapid development of cloud computing, has been laid a foundation of cloud storage from its concept to the actual ground. At the same time, due to the development of terminal technology and mobile network rapid popularization, the application range of cloud storage has also been enlarged. The model of the Cloud plus the terminal is more closed to people’s lives, and it is changing the people’s daily habits.In this paper, the cloud storage technology has been studied, and cloud storage system solution was proposed in view of the terminal application. For the small document type two different ways of storing a copy were adopted:one is the relational database MySQL; the other is a non-relational database MongoDB. For the large file type, the dynamic consistency hash algorithm was used to store them. At the same time the particle swarm optimization algorithm was also introduced to select the optimal storage node for cloud storage data. The performance of the system was tested through small scale actual operation the system and the cloudsim simulation. The main work was described as followings:(1)Through reviewing the journals and electronic resources, the source codes in the open source and distributed file system, such as the HDFS and FastDFS, were researched. Based on the analysis of system’s functional requirements and non-functional requirements, the architecture of cloud storage system was designed.(2)The main functions of the client and server program in the system was developed in details, and in ordinary PCS a small cloud storage system was programmed and built. Through the cloud storage system, the copy strategy put forward in the following paragraphs was verified, and the particle swarm optimization algorithm to be applied in the optimal storage node selection of the data copy was also be verified. In the process of system design, considering the characteristic of mobile terminals, geographically the nearest cloud storage service was used.(3)The copy strategy in the system was researched. In this paper, the classification processing method was used:for the small files the database approach was adopted, and for the large files dynamic consistency hash algorithm was introduced. Database approaches also were used two kinds strategies to analyze and design:one is the relational database MySQL scheme; the other is the non-relational database MongoDB schema. Large file type was designed as a decentralized structure to eliminate performance bottlenecks in the process of the concurrent processing. Meanwhile the copy was combined with the particle swarm optimization algorithm.(4)In the data replication optimal storage node selection process the particle swarm optimization algorithm is introduced. The node selection is in discrete space, particle swarm optimization algorithm was redefined in the discrete problem space. What is more, the fitness evaluation function of the problem was defined to evaluate the solutions. Meanwhile, in order to get a more optimal solution, the method of the heuristic information was introduced to guide the algorithm for better solutions.