计算机应用 ›› 2011, Vol. 31 ›› Issue (05): 1400-1403.DOI: 10.3724/SP.J.1087.2011.01400

• 数据库技术 • 上一篇    下一篇

闪存数据库系统中一种高效的自适应存储模式

王立1,王跃清1,王翰虎1,2,陈梅1   

  1. 1.贵州大学 计算机科学与信息学院,贵阳 550025
    2.贵州星辰科技开发有限公司,贵阳 550025
  • 收稿日期:2010-11-23 修回日期:2011-01-15 发布日期:2011-05-01 出版日期:2011-05-01
  • 通讯作者: 王立
  • 作者简介:王立(1985-),男,陕西西安人,硕士研究生,CCF会员,主要研究方向:闪存数据库系统;王跃清(1985-),男,内蒙古呼和浩特人,硕士研究生,CCF会员,主要研究方向:闪存数据库系统;王翰虎(1946-),男,贵州遵义人,教授,CCF高级会员,主要研究方向:分布式系统、数据库系统;陈梅(1964-),女,贵州贵阳人,教授,主要研究方向:数据库、软件工程。
  • 基金资助:

    贵阳市科技攻关项目([2010]筑科工合同字第28号);贵州大学2010年研究生创新基金资助项目(校研理工[2010033])。

Efficient and self-adaptive storage schema in flash-based database system

WANG Li1, WANG Yue-qing1, WANG Han-hu1,2, CHEN Mei1   

  1. 1.School of Computer Science and Information, Guizhou University, Guiyang Guizhou 550025, China
    2.Guizhou XingChen Technology Development Company Limited, Guiyang Guizhou 550025, China
  • Received:2010-11-23 Revised:2011-01-15 Online:2011-05-01 Published:2011-05-01
  • Contact: Li Wang
  • Supported by:

    Industrial research projects Fund of Guizhou Province;Innovation Fund of Guizhou University;Industrial research projects Fund of Guizhou Province

摘要: 使用闪存作为存储介质成为提高数据库系统性能的一条新途径,为了解决闪存数据库系统存储管理技术中基于日志的更新策略存在查询效率低、日志区空间分配不合理、索引更新代价高等问题,提出了基于Bloom Filter的最新版本预测算法,引入记录定位器结构,提出日志概要结构和基于闪存更新查询代价评估模型的自适应机制。实验证明,该方法能够自适应地划分合理的日志区空间,有效提高查询性能,减少各种非聚集索引的更新代价。

关键词: 闪存, 数据库, 存储管理, 自适应机制, 代价评估

Abstract: It has been a new approach that flash memory is used as storage media to improve database system performance. In the current flash database system, to solve the problems of low search performance, improper log region allocation, and high index update cost in storage management, a forecasting algorithm based on the latest version of Bloom Filter was proposed, record locator and log summary structure were introduced, and a self-adaptive mechanism based on flash search and update cost estimate model were given. The experimental results prove that it can make a proper log region allocation, efficiently improve search performance, and reduce non-clustered index update cost.

Key words: flash, database, storage management, self-adaptive mechanism, cost estimate