RMI-DRE： a redundancy-maximizing identification scheme for data redundancy elimination

• Published in: Science China. Information sciences 2016-08, Vol.59 (8), p.252-254, Article 089301
• Main Authors: Zhang, Nan, Yang, Xiaolong, Zhang, Min, Long, Keping
• Format: Article
• Language: English
• Subjects: Computer Science; Information Systems and Communication Service; IP流量; Letter; MRC; Redundancy; 匹配机制; 块匹配; 数据冗余; 数据包; 最大化; 识别
• ISSN: 1674-733X; 1869-1919
• DOI: 10.1007/s11432-016-5523-y

Dear editor, Data redundancy elimination （DRE） technology can efficiently reduce the redundant IP traffic in the network [1,2]. A detailed DRE process consists of many steps, such as the fingerprint selection, chunk matching, packet encoding, and packet decoding [3]. Fingerprint selection is one of the most important factors for DRE. In traditional DRE, the redundant chunk size is set to a fixed value. The fixed chunk size will affect the DRE performance seriously, so keeping the redundant chunk size variable is necessary [4]. Hence, multi-resolution chunking （MRC） has been proposed. However, MRC does not consider the overlap between the adjacent chunks, which in- evitably brings some unnecessary overhead. Be- sides, chunk-matching is another key point for DRE. There are two popular chunk matching mechanisms, i.e., Chunk-Match and Max-Match. Chunk-Match may not bring much storage over- head, but the size of chunk-matching is lim- ited, which results in some redundant bytes to be missed. By contrast, Max-Match can maxi- mize the size of chunk-matching by extending the boundary of the identified redundant chunk to- wards its front or rear.