并行数据库查询优化技术研究  

Query optimization technique for parallel databases

作　　者：徐丽萍[1] 金雄兵[1] 赵小松[1] 

机构地区：[1]华中科技大学计算机科学与技术学院,湖北武汉430074

出　　处：《华中科技大学学报（自然科学版）》2006年第3期11-13,20,共4页Journal of Huazhong University of Science and Technology(Natural Science Edition)

摘　　要：并行数据库查询优化的关键是缩减庞大的计划搜索空间,针对无共享结构(SN),充分考虑通信开销,给出代价估计模型,采用两阶段优化方法,依据代价估计模型先对查询树进行顺序优化,并提出一种新的两阶段查询优化策略.对顺序优化的查询计划进行并行化,充分利用了多处理机的并行性,获得了较快的查询响应速度,既具备良好的并行性又避免了复杂性.The aim of the study is to prune the big search space to optimize query techniques for parallel databases. Taking communication costs into account, a cost-evaluated model based on shared-nothing structure was proposed. With the model, two-phased optimization strategy was used to optimize the query tree sequentially, and the sequential optimization tree was parallelized in the second phase. After parallelization of multi-processors was used, faster response-time was acquired. The good parallelization was achieved and the complexity was avoided by the new strategy.

关 键 词：并行数据库 查询优化 两阶段优化 

分 类 号：TP311.5[自动化与计算机技术—计算机软件与理论]