HCSIM一种长期高频Block-Level快照索引技术.pdfVIP

第32卷第10期 计 算 机 学 报 v01．32 No．10 2009年10月 CHINESE JOURNAL OF C()MPUTERS Oct．2009 HCSIM ●-一种长期高频Block-Level快照索引技术 吴广君 云晓春2’ 方滨 兴”’2’ 王树鹏2’ 余翔湛” I’(哈尔滨工业大学计算机网络与信息安全技术研究中心哈尔滨150001) 2’(中国科学院计算技术研究所北京 100190) 摘要高频快照技术应用于备份时，能够为物理错误和人为错误提供数据保护，构建可靠数据存储环境．针对长 期、高频block—level快照检索效率低下问题，在对目前常见的block-level快照技术建模、分析基础上，提出结合数 据分布特征和检索模式的分层次二维索引结构——HCSIM．实验、分析结果表明：HCSIM索引技术应用于长期、高 频快照管理时，比时空索引结构Overlapping B+Tree索引技术显著提高索引的存储效率和检索效率；通过定性的 分析．HCSIM是目前存在的block-level快照索引模式中存储效率和检索效率相对平衡的索引技术． 关键词快照；安全存储；索引；灾难恢复；块数据 中图法分类号TP393 DOI号：10．3724／SP．J．1016．2009．02080 HCSIM：An Indexing Method for Long—Lived Frequent Block—Level Snapshot WU Guang—Jun¨ YUN Xiao—Chun∞ FANG Bin—Xin91’孙 WANG Shu—Pen92’ YU Xiang—Zhanl’ 11(Research Center of Computer Network and Information Security Technology，Harbin Institute of Technology，Harbin 150001) 2’(Institute o／Computing Technology。Chinese Academy of Sciences。Beijing 100190) Abstract Snapshot—based backup techniques can protect data from accidental and physical errors and support reliable storage．The long—lived frequent snapshot is low in search efficiency，because of the shared data between versions．Up to now there is no efficient indexing method for the long— lived frequent block—level snapshot．Adapting the block—level snapshot distributing features，this paper proposes a general—purpose tWO—dimensional indexing structure：Hierarchical Clustering Snapshot Indexing Method(HCSIM)．The experimental results show that HCSIM can dramatic— ally increase indexing storage efficiency than traditional temporal and spatial indexing structure Overlapping B+Tree．At the meantime，HCSIM can achieve better query efficiency in the con— text of long—lived snapshot management．The theoretical analysis exposes that HCSIM structure can achieve better balance between storage and query efficiency than nowadays block——level snap·· shot indexing method． Keywords snapshot；rel iable storage；indexing；disaster recovery；block—l evel 收稿日期：2009—07—15；最终修改稿收到日期：2009～08—21．本课题得到国家“八六三”高技术研究发