云环境下元数据弹性分级一致性保障机制研究.pdfVIP

Vol. 33 No.7 第 33 卷第 7 期 计算机应用研究 2016 年7 月 Application Research of Computers Jul. 2016 云环境下元数据弹性分级一致性保障机制研究* 丁海骏，卢菁 (上海理工大学光电信息与计算机工程学院，上海2α093 ) 摘 要:为了解决元数据的动态更新导致的元数据副本的不一致，从云存储系统中不同元数据对一致性强度需 求的差异性出发，提出了基于元数据弹性分级的一致性保障机制。根据元数据访问频率以及更新属性敏感度对 元数据更新弹性分级，结合基于读优化的 Quorum 算法实现元数据更新的分级一致性保障。实验证明该机制可 以有效权衡元数据副本可用性、一致性和系统开销，相比单一的一致性策略更为灵活。 关键词:云存储;元数据;副本;一致性;弹性分级 中国分类号:T凹91 文献标志码:A 文章编号: 1∞1-3695(2016)07-2039-04 doi: 10. 3969/j. issn. 1 ∞1-3695.2016.07.026 Elastic classification based consistency service for cloud storage Ding Haijun , Lu Jing ( School o/Optical-Electrical Co叩uter Engineering , University 0/ S加nghai/or sc陀阳 Technology , Shanghai 2()(朋3 ， China) Abstract: In order ωsolve the inconsistency of metadata made by dynamic metadata update ,according to different require- ments of metadata consistency strength for each metadata ,this paper presented a consistency service based on elastic classifica- tion of metadata for cloud storage. Elastic classification of metadata was based on metadata access frequency and update attri- bute sensitivity ,it combined with read optimization based Quorum algorithm to realize classified metadata consistency guaran- tee. Experiments show that the mechanism can effectively ba1ance the metadata replica availability ,consistency and system 0- verhead. Compared with the single strength consistency strategy ,it is more flexible. Key words: cloud stora在町 metadata; replica; consistency; elastic classification 统的数据副本一致性策略对于系统性能有着直接的影响。在 0 号|言