摘要
研究分布式数据库节点准确选择问题。大型分布式数据库系统一旦遇到突发故障,可以从一个出现故障的副本节点切换到另一个正常运行的节点,不会因为某个节点的故障导致系统的瘫痪,传统的节点切换选择以功能近似性为基础,选择功能近似的节点进行替换,但是没有考虑节点本身的自带属性,对节点考察只从功能相似性考虑,很容易选出坏点或者死亡节点,造成数据库通信可靠性下降。提出一种基于选择回归树算法的分布式数据库同步节点选择优化方法。根据粗糙集相关理论,对分布式数据库中的海量节点进行属性权重计算。根据节点的属性权重,利用选择回归树方法,完成分布式数据库同步节点选择优化处理。实验结果表明,运用改进算法进行分布式数据库节点选择,可以极大的提高分布式数据库的可靠性。
The accuracy of distributed database node selection was researched. In this paper, distributed data-base synchronous node optimization selection method based on a regression tree algorithm was proposed. On the basis of rough set theory, attribute weights for the mass nodes in distributed database were calculated. According to the node attribute weights, the selection regression tree method was applied to complete optimize the synchronize distributed database node selection. The experimental results show that the presented algorithm used in distributed database node selection can greatly improve the reliability of distributed database.
出处
《计算机仿真》
CSCD
北大核心
2013年第12期407-410,共4页
Computer Simulation
关键词
分布式数据库
节点选择
粗糙集
选择回归树
Distributed database
Node selection
Rough set
Choose regression tree
