为了使列存储OLAP(on-line analytical processing)操作中I/O和CPU开销较大的扫描、连接、聚集操作实现有效的共享和复用,提出了一个多查询优化技术。根据列存储以及OLAP操作的特点,提出了一系列转换规则,为OLAP查询请求产生的一组相关...为了使列存储OLAP(on-line analytical processing)操作中I/O和CPU开销较大的扫描、连接、聚集操作实现有效的共享和复用,提出了一个多查询优化技术。根据列存储以及OLAP操作的特点,提出了一系列转换规则,为OLAP查询请求产生的一组相关查询语句生成一个单一全局查询计划。为了达到共享复用的目的,在全局计划中引入新的过滤结点、分组结点、合并结点和聚集结点。同时,借用MuGA(multiply group by algo-rithm)算法,通过分组结点、合并结点、连接结点实现维表及事实表元组的分组序号标记,从而实现列扫描、列连接的共享。并为聚集结点提出了一个多阶段聚集算法,结合最终生成的事实表复合分组序号,实现聚集操作的复用。在SSB(star schema benchmark)数据集上设计实验,证明了该多查询优化策略的有效性。展开更多
In this paper, we introduce the origination of the DW ideology and its architecture & principles,as well as relative techniques. With the example that the MIS of ENWEI company, we discuss how to apply the base the...In this paper, we introduce the origination of the DW ideology and its architecture & principles,as well as relative techniques. With the example that the MIS of ENWEI company, we discuss how to apply the base theories of DW and implement its main techniques in modern companies. At the end of this paper ,we propose the ways that how to build an effective Decision Support System based on the MIS of ENWEI company.展开更多
文摘为了使列存储OLAP(on-line analytical processing)操作中I/O和CPU开销较大的扫描、连接、聚集操作实现有效的共享和复用,提出了一个多查询优化技术。根据列存储以及OLAP操作的特点,提出了一系列转换规则,为OLAP查询请求产生的一组相关查询语句生成一个单一全局查询计划。为了达到共享复用的目的,在全局计划中引入新的过滤结点、分组结点、合并结点和聚集结点。同时,借用MuGA(multiply group by algo-rithm)算法,通过分组结点、合并结点、连接结点实现维表及事实表元组的分组序号标记,从而实现列扫描、列连接的共享。并为聚集结点提出了一个多阶段聚集算法,结合最终生成的事实表复合分组序号,实现聚集操作的复用。在SSB(star schema benchmark)数据集上设计实验,证明了该多查询优化策略的有效性。
文摘In this paper, we introduce the origination of the DW ideology and its architecture & principles,as well as relative techniques. With the example that the MIS of ENWEI company, we discuss how to apply the base theories of DW and implement its main techniques in modern companies. At the end of this paper ,we propose the ways that how to build an effective Decision Support System based on the MIS of ENWEI company.
