摘要
分析了数据划分模式对并行数据库系统整体性能的影响,设计了并行数据库系统中数据自动划分的框架,开发了基于DB2查询优化的交互式划分工具. 应用该划分工具对一个给定的由一组SQL语句组成的负载,可寻找在多个节点间划分数据的有利模式. 测试结果表明:该交互工具能够在合理的时间内为负载中每个语句推荐好的候选划分,并可避免初始数据倾斜,减轻数据库管理员的负担.
Through declustering the tuples of relations on disks of many nodes and uses the I/O parallelism, the parallel database systems obtain a higher performance. Data Partition directly affects the parallel database system's overall performance and scalability. This paper attempts to automate data partition for PDBMS. We have established a partition tool based on the DB2 inquiry optimization program, which can release the burden of DBA and avoid the initial data skew.
出处
《大庆石油学院学报》
CAS
北大核心
2005年第2期73-75,127,共4页
Journal of Daqing Petroleum Institute
