| In current information society, data is the most valuable asset of an enterprise or organization, the damage of crucial data will cause fatal strike to enterprise. In the mean time, data is very vulnerable in face of natural disaster, man-made misoperations, virus and hacker attacks. Data backup, an effective means of data protection, is receiving more and more attention and application. With the popularity of cloud computing and SaaS (Software as a Service), backup service gradually substitutes traditional backup software, and becoming a new research direction in storage application area.Currently, backup software and service market is monopolized by foreign products, it is of great significance to guarantee the information security in china by means of studying key technologies and developing backup service system with independent intellectual property rights.Aiming at the data transmission and storage load balance of concurrent data streams transferring between large scale users and multiple medium nodes, the paper investigated the concurrency scheduling technology conducted by director server. In the macroscopic level, real-time scheduling strategy and whole time scheduling strategy are proposed, which are used to connect user and the appropriate storage node to accomplish data transmission, both of the two strategies and their adaptabilities are analyzed and compared; in the microscopic level, multithreading management mechanism is studied, towards the currently used ad-hoc mode, a improved thread pool approach is presented.In order to improve the efficiency of massive data transportation and storage, this paper investigated the data organization in the storage server and the data compression technology covered both backup client and storage server. In terms of data organization, the storage files across nodes are independent. Files within a node can grow dynamically according to users' requirement. In each storage file, datum are stored in data chunk, different data chunks are indexed by their hash values. In terms of data compression, a hybrid data compression approach is proposed, which is composed of two levels: the lower level is block compression, during the course of data backup, to the data block inexistent in storage server, it will be compressed by backup client before transferring to the storage server, so as to eliminate the redundant data stream within a single file; the higher level is global compression, in the process of data backup, if a data chunk has been existent in the storage server, then the same copy doesn't need to be transferred repeatedly, server only need to modify the data index, so as to accomplish de-duplication across multiple files from different users. In the end, three block compression algorithms are adopted to test their compression ratios and CPU utilization ratios respectively.Since there are more difficult technology problems in the online backup service system than that in backup software, which more prone to exceptions objectively, at the same time, higher service dependability is required subjectively, so the paper studied the dependability of online backup service. First, dependability notion and dependability analysis and improvement theory is investigated, then Fault-tolerant Online Backup Service Model (FOBSM) is proposed, and formal language Object-Z is used to build formal models for the components of FOBSM, then reasoning rules are followed to demonstrate the fault-tolerant properties of FOBSM, this research has valuable guidance to the fault-tolerant design for the online backup service system.On the basis of the above research, an online backup service prototype system was designed and implemented.
|
PDF Full Text Request