欢迎访问中国科学院大学学报,今天是

中国科学院大学学报 ›› 2012, Vol. 29 ›› Issue (1): 94-100.DOI: 10.7523/j.issn.2095-6134.2012.1.014

• 计算机科学 • 上一篇    下一篇

集群数据库系统多指标动态负载均衡方法的设计与实现

殷佳欣, 陈驰   

  1. 中国科学院软件研究所信息安全国家重点实验室, 北京 100190
  • 收稿日期:2010-08-16 修回日期:2010-12-30 发布日期:2012-01-15
  • 基金资助:

    国家863计划项目(2007AA120404,2007AA120405)、国家自然科学基金(61003228)和中国科学院知识创新工程领域前沿项目(ISCAS2009-DR13)资助

A multi-criteria dynamic load balancing method for clustered databases: design and implementation

YIN Jia-Xin, CHEN Chi   

  1. State Key Laboratory of Information Security, Institute of Software, Chinese Academy of Sciences, Beijing 100190, China
  • Received:2010-08-16 Revised:2010-12-30 Published:2012-01-15

摘要:

负载均衡是提高数据库集群系统整体性能的重要技术手段之一,其核心在于如何对实际到达的任务进行合理的动态分配.在对比分析现有数据库集群系统的负载均衡方法的基础上,提出一种新的多指标动态负载均衡方法.该方法对后端数据库节点的多个性能指标进行采样,使用多指标决策方法TOPSIS计算各个节点负载情况,用负载最轻的节点作为决策结果.实验结果表明:与传统负载均衡方法比较,该方法可显著提高集群数据库系统的整体性能.

关键词: 集群数据库, 负载均衡, 多指标决策, 负载检测

Abstract:

Load balancing is one of the most important technical methods for improving the overall performance of clustered databases. On the basis of comparisons of several present load balancing methods, we present a multi-criteria dynamic load balancing method.The method was designed to collect load state at all the backends of the cluster, make an ordered sequence of the backends using TOPSIS multi-criteria decision method, and send the request to the backend supposed to be the lightest in burden. On the basis of the experimental results we come to the conclusion that the method does improve the performance of the clustered database system compared to traditional load balancing methods.

Key words: clustered database, load balancing, multi-criteria decision, load detect

中图分类号: