Research On Network Differential Backup System Based On Identical Data Detection

With the development of information technology，the importance of data forenterprises becomes more prominent. It is really effective for data protection to backupdata to network storage system. With the rapid growth of corporate data, the backup ofthese massive data brings more pressure to storage center and network resource, andalso leads the storage center to incur a lot of redundant data. So how to eliminateredundant data and save storage space and network bandwidth becomes the key issuesthat need to be solved urgently. Identical data detection technique appears to be aneffective technique to reduce the amount of physical storage of backup data andnetwork transmission.The detection accuracy and performance of existing identical data detectionalgorithms are quite different because of different data detection size. Block-levelidentical data detection can balance detection accuracy and efficiency, but it still can beimproved on detection accuracy. Based on the present research of identical datadetection technique, this thesis proposes Sliding Block with Backtrace Sub-blockalgorithm for higher detection accuracy, and designs a network differential backupsystem based on identical data detection technique. The main contributions of this thesisare as follows:①Describe the principle of the three typical block-level identical data detectionalgorithms: Fixed-sized Partition, Variable-sized Partition and Sliding Block, andanalyze the advantages and disadvantages of these three algorithms.②Aiming at the deficiency of Sliding Block, this thesis proposes Sliding Blockwith Backtrace Sub-block algorithm. Experimental results show that this method canimprove detection accuracy about6.5%than Sliding Block, and16.5%thanContent-defined Chunking. According to the results of experimentation, analyze theinfluence of data block dividing method on detection accuracy and extra spaceoverhead.③Design Network Backup Restore Protocol(NBR) to meet communication needsof network backup. NBR is above the Transport Layer, and supports encryption,compression and transmission of packets, and can meet the communication and filetransfer needs.④According to the above research work, this thesis designs a network differential backup system based on identical data detection. Main work is included as follows: thisthesis designs and implements the backup client based on identical data detection, themain modules of server which are central control module and backup and restoreresponse module, and the common network communication module of server and clientbased on NBR protocol.The experimental results show that our designed system can significantly reducethe network bandwidth and save storage space when compared with Sliding Blockmethod in terms of differential backup, as well as support multiple users in backupingor restoring concurrently and exhibit significantly practical value.