Reseach On Data Deduplication Technology In Cloud Backup For Education System

The society has been informationalized and digitalized in speed in 21 st century. The amount of data generated by people everyday is immeasurable. Enormous physical disks and network resources were consumed during the production of data and the cost of memory management is in challenge. In addition, duplications of different users’ data are stored in the same memory system in the concept of cloud backup. Thereby, the optimization of data storage space has become a hot area of research.Researchers found a large proportion of data duplicate in various kinds of current data storage systems. De-duplication technology was proposed to optimize the physical space of data storage. De-duplication is also the focus of storage area in current years. The essence of de-duplication technology is only one duplicate data block in storage system has a real copy, and each of other blocks merely has a pointer to the copy. Thus, the utilization percent of physical memory resource can be improved. This thesis presents an optimal fingerprint filtering index strategy in accordance with the development and requirements of de-duplication technology. The basic component of the strategy is a counter-type bloom filter, which supports the extension of data index and the removal of data fingerprint. Then, the thesis designs an educational cloud backup system based on de-duplication and implements the functions of each module of the system. The educational cloud backup system upgrades the scalability of the storage space by virtualizing physical resources without letting users know the distribution details. The de-duplication technology could increase the storage utilization of the educational cloud backup system while shrinking the cost of transferring duplicate data.In the end of the thesis first tests the functions of the educational cloud backup system. The result indicates the availability of the system. Then the actual utilization of physical memory space is tested, which turns out to be highly increased. At last, an experiment to examine the performance of the fingerprint filtering index strategy is also conducted successfully.