Research On Data Deduplication Based On File Access Patterns

With mobile devices growing, data synchronization，sharing and protection betweenequipments are more important. Therefore, in recent years personal cloud storage servicesare provied by major companies. Apparently synchronization faster, smaller storage space,the easier it is to let the user satisfaction. The data compression has a very important rolefor such systems: for the user it can reduce the amount of uploaded data, so fastersynchronization; for service providers, it can reduce the storage space and cost.Traditional storage capacity optimization techniques, such as traditional losslesscompression algorithm, delta compression algorithm have their own limitations. They cannot provide a high compression ratio. Deduplication technology is a new kind of storagecapacity optimization technology to identify and remove redundant data, and will replacethe redundant data with a space efficient index. The performance bottlenecks, such ascalculating and the index bottlenecks, introduce the problems in terms of data reliabilityand readability problems.Deduplication is provided based on file access patterns. Classification according tothe behavior of the file access can help remove redundant data of different file classes byusing different strategies, the modified file using delta compression in user space, thenewly created or copied from other using global file-level Deduplication across users. Asynchronization module with related strategies is implemented. The experimental resultsshow that this method can reduce user actual synchronization time and data storagecapacity, and allows the overhead within the acceptable level.