计算机工程

• 先进计算与数据处理 • 上一篇    下一篇

固态硬盘混合存储数据库的数据分布优化算法

周世民1a,1b ,柴云鹏1a,1b ,王 良1b ,王 鑫2   

  1. (1. 中国人民大学a. 数据工程与知识工程教育部重点实验室; b. 信息学院,北京100872;2. 天津大学计算机科学与技术学院,天津300072)
  • 收稿日期:2014-07-22 出版日期:2015-04-15 发布日期:2015-04-15
  • 作者简介:周世民(1989 - ),男,硕士,主研方向:数据库技术;柴云鹏(通讯作者),讲师、博士;王 良、王 鑫,副教授、博士。
  • 基金项目:

    国家“863”计划基金资助项目(2013AA013204);国家自然科学基金青年基金资助项目(61202115);计算机体系结构国家重点实验室开放课题基金资助项目(CARCH201302)。

Data Layout Optimization Algorithm for Database of Hybrid Storage with Solid State Drive

  1. (1a. Key Laboratory of Data Engineering and Knowledge Engineering,Ministry of Education; 1b. School of Information,Renmin University of China,Beijing 100872,China; 2. School of Computer Science and Technology,Tianjin University,Tianjin 300072,China)
  • Received:2014-07-22 Online:2015-04-15 Published:2015-04-15

摘要:

基于闪存的固态硬盘(SSD)可以有效提升联机事务处理(OLTP)数据库的性能,但由于目前SSD 价格仍然较高,一般多与磁盘组成混合存储。为此,提出数据分布的自适应优化算法以及具体的优化策略。该算法能够自动适应应用的特征,通过观测判断各个数据元素的性能提升效率,从而在SSD 和磁盘之间自动形成理想的数据分布。基于实际数据库系统的实验结果表明,该算法可适应各种SSD 空间配置,使基于混合存储的OLTP 数据性能得到有效提升。

关键词: 闪存, 混合存储, 固态硬盘, 数据库, 联机事务处理, 自适应, TPC-C 测试

Abstract:

Flash-based Solid State Drive ( SSD ) can improve the performance of On-line Transaction Processing (OLTP) database efficiently. Due to the high cost of SSD products,however,SSD is usually utilized in hybrid storage combined with traditional disks. Therefore,this paper proposes an adaptive data layout optimization algorithm and two specific strategies. The algorithm can adjust the characteristic of application adaptively,through observing and deciding the performance improvement efficiency of each data element,to form an optimized data layout between SSD and disks. Experimental results based on a practical database system show that the algorithm is flexible and efficient to adapt to various SSD capacity configurations,and can improve the performance of the OLTP database based on hybrid storage effectively.

Key words: flash, hybrid storage, Solid State Drive (SSD), database, On-line Transaction Processing (OLTP), selfadaption, TPC-C test

中图分类号: