| Cloud storage is the development and extension of the concept of cloud computing, and it has the features of large capacity, high performance, low cost, high versatility and ease of scalability. Its emergence and development has changed the traditional data storage, and it can provide high-quality solution for data storage and management. It is becoming an important choice for a lot of business users and individual users to improve efficiency and cost less.In a cloud storage system, replication technology is a common technique for improving its reliability and availability of the system, and it has an important role on reducing the system response time. A strategy of replication technology includes creation, deletation, location, and consistency maintenance. In order to improve the availability and performance of the system, this paper research on the HDFS platform, and the main research points are the redundancy settings, and layout strategy. The main work includes:1、The paper studied the research status and key technology of cloud storage. Base on the HDFS framework, the paper focused on the replication technology of cloud storage and summarized its characteristics and problems of management analysis.2、For the shortcomings of the redundancy in HDFS, we proposed a mechanism based on the heat of documents,which it can be dynamically defined by the frequency of file accessed. This paper described specific how to calculate the heat of document, as well as to determine the redundancy through the heat, and described the process to create a copy.3、In the replication layout of HDFS, on the basis of perception of the rack, we proposed a strategy based on the classification of user data and improved TOPSIS algorithm. In the strategy, we classified the data based on user demand, and improved the TOPSIS algorithm based on gray correlation coefficient optimal node, and then sorted the nodes through the improved TOPSIS algorithm, and last, chose the apposite node for the users. |
PDF Full Text Request