基于分组序号的聚集算法  

An Aggregation Algorithm Based on Group Numbers

作　　者：冯建华[1] 蒋旭东[1] 孟宪虎[2] 

机构地区：[1]清华大学计算机科学与技术系,北京100084 [2]运城高等专科学校计算机系,山西运城044000

出　　处：《软件学报》2003年第2期222-229,共8页Journal of Software

基　　金：国家重点基础研究发展规划(973)~~

摘　　要：联机分析处理OLAP(online analytical processing)查询作为一种复杂查询,当使用SQL(structured query language)语句来表述时,通常都包含多表连接和分组聚集操作,因此提高多表连接和分组聚集计算的性能就成为ROLAP(relational OLAP)查询处理的关键问题.提出一种基于分组序号的聚集算法MuGA(group number based aggregation with multi-table join),该方法充分考虑数据仓库星型模式的特点,将聚集操作和新的多表连接算法MJoin(multi-table join)相结合,使用分组序号进行分组聚集计算,代替通常的排序或者哈希计算,从而有效地减少CPU运算以及磁盘存取的开销.算法的实验数据表明,提出的MuGA算法与传统的关系数据库聚集查询处理方法以及改进后的基于排序的聚集算法相比,性能都有显著提高.OLAP (online analytical processing) queries are complex. When implemented in SQL (structured query language), they usually involve multi-table join and aggregate operations. As a result, how to improve the performance of the multi-table join and aggregate operations becomes a key issue for ROLAP (relational OLAP) query evaluation. To solve this problem, an aggregation algorithm based on group numbers named MuGA (group number based aggregation with multi-table join) is proposed in this paper. By taking the characteristics of star schema into consideration, the algorithm combines the aggregation operation with the novel multi-table join algorithm, Mjoin (multi-table join), and replaces the sorting and hashing method by computed group numbers in aggregation computing. As a result, the algorithm can not only reduce the CPU time, but also reduce the disk I/Os for OLAP queries. As illustrated by the experiments, the performance of the algorithm MuGA is superior to original aggregation methods and the new sorting based method for aggregation.

关 键 词：分组序号 聚集算法 数据仓库系统 关系数据库 

分 类 号：TP311.32[自动化与计算机技术—计算机软件与理论]