Design And Implementation Of Data Synchronism, Backup And Recovery In Private Cloud

Private cloud is currently a hot direction in the field of cloud computing. To enable the user to work on the cloud, more and more enterprises and companies build a private cloud environment in their interior. While cloud storage is the most important part of the private cloud, how to ensure the data service’s reliability and the security of the data becomes one of the most important problem in the private cloud environment. However, data storage scheme, provided by the cloud service providers, is an expensively cost. Except for that, the fault tolerance of the existing scheme is very poor and not suitable for the small and medium-sized enterprises. To deal with all the above issues, in this thesis, we propose a novel data storage scheme, which is open source and reliable in the private cloud.Firstly, in this thesis, we carry on a comprehensive analysis and summary for the current mainstream of data storage scheme and key technologies of reliable storage system. On this basis, we design and implement a system, which is faculty of data synchronization, backup and recovery that includes management module, data control module and the LVS(Linux Virtual Server) cluster module. Meanwhile, we design an interface to apply for communication between modules. At first, in management module, by the combination of the heartbeat detection and floating IP technology, we implement the functions of automatic detection and processing fault. Then, considering the convenience of system management, we design the visualization of the management interface on the module structures. Furthermore, in order to enhance the reliability of the storage system, the data control module is divided into master node, slave node and backup node, and these three nodes implement the functions of the data synchronization, data backup and restore, respectively. For data synchronization, we copy data that will be written into the master node in real time, to the slave node using the hard disk data interception technology. Thus, we can guarantee the data consistency of master-slave node. Guaranting the rational use of system resources and enhancing system performance, we adopt two management modes,manual mode and planning mode, when implement system backup and restore functions. On the consideration of reducing the overall data transmission time, we decide to apply the task model based on epoll and thread pool, rather than the traditional pattern of multi-process copy data, to implement the function of data coping. In the LVS cluster module, we apply the i SCSI storage technology to provide the service of the users’ data access that makes the system extensible.Finally, we evaluate the system from the perspective of function and performance. The results demonstrate that the system meets the basic requirement of reliable storage system with limited system resources. Furthermore, the prototype has a higher performance and proves the feasibility of the system in a private cloud.