基于并发跳表的云数据处理双层索引架构研究-论文.pdf

《基于并发跳表的云数据处理双层索引架构研究-论文.pdf》由会员上传分享，免费在线阅读，更多相关内容在行业资料-天天文库。

1、计算机研究与发展DOI：10．7544／issnl000—1239．2015．20i40358JournalofComputerResearchandDevelopment52(7)：1531-1545，2O15基于并发跳表的云数据处理双层索引架构研究周维路劲周可人王世普姚绍文(云南大学软件学院昆明650091)(zwei@ynu．edu．cn)ConcurrentSkiplistBasedDouble-LayerIndexFrameworkforCloudDataProcessingZhouWei，LuJin，ZhouKeren，WangShipu，

2、andYaoShaowen(SchoolofSoftware，YunnanUniversity，Kunming650091)AbstractClouddataprocessingplaysanessentialinfrastructureincloudsystems．Withoutefficientstructures，cloudsystemscannotsupportthenecessaryhighthroughputandprovideservicesformillionsofusers．However，mostexistingcloudstora

3、gesystemsgenerallyadoptadistributedHashtable(DHT)approachtoindexdata，whichlackstosupportrange—queryanddynamicreal—timecharacter．Itisnecessarytogenerateascalable，dynamicalandmulti—queryfunctionalindexstructureincloudenvironment．Basedonthesummaryandanalysisofthedouble—layerindexsy

4、stemsforcloudstorage，thispaperprovidesanovelconcurrentskiplistbaseddouble—layerindex(referredasCSD-index)forclouddataprocessing．Two—layerarchitecture，whichcanbreakthroughsinglemachinememoryandharddrivelimitation，isusedtoextendindexingscope．Onlinemigrationalgorithmofskiplist’Snod

5、esbetweenlocalserversisusedtomakedynamicload—balancing．Thedetailsofthedesignandtheimplementoftheconcurrentskiplistarediscussedinthispaper．Optimisticconcurrencycontrol(OCC)techniqueisintroducedtoenhancetheconcurrency．ThroughconcurrentskiplistCSD—indeximprovestheloadbearingcapacit

6、yoftheglobalindexandenhancestheoverallthroughputoftheindex．Experimentalresultsshowtheefficimmyoftheconcurrentskiplistbaseddouble—layerindexandithasviabilityasanalternativeapproachforcloud—suitabledatastructures．Keywordscloudcomputing；double-layerindex；concurrentskiplist；rangeque

7、ry；optimisticconcurrencycontro1摘要云数据处理在云计算基础设施中占有极其关键的地位．然而，当前的云存储系统绝大部分都采用基于分布式Hash的健一值对模式来组织数据，在范围查询方面支持不理想、且动态实时性差，有必要构建云环境下辅助动态索引．通过总结、分析云环境中辅助双层索引机制，提出一种基于并发跳表的云数据处理双层索引架构．该架构采用两层体系结构，突破单台机器内存和硬盘的限制，从而扩展系统整体的索引范围．通过动态分裂算法解决局部服务器中的热点问题，保证索引结构整体的负载均衡．通过并发跳表来提高全局索引的承载性能，改善了全局索

8、引的并发性，提高整体索引的吞吐率．实验结果表明，基于并发跳表的云数据处理双层索引架构能够有效支