摘要
OLAP系统通过对企业数据的智能查询,可以对企业决策起到良好的辅助作用。但是大量复杂的查询分析任务要求OLAP系统部署在高性能的服务器上,从而使得系统成本高昂,限制了其在广大中小企业中的推广应用。通过引入分布式技术,把OLAP系统部署在多个小型服务器上,把大量的查询分析任务分配到各个服务器上分布执行,可以利用现有的企业资源来实现OLAP应用。分布式系统中,系统的协调者一般是通过选举算法选出的,在全连接的局域网内,选举策略一般采用Garcia-Molina提出的经典欺负算法或相应的改进算法。本文结合分布式OLAP系统的特点,提出了一种适合该系统的改进选举算法,减少了通讯量,提高了系统效率。
The OLAP system can support enterprise decision well through intelligent query with data of enterprise. To deal with abundant and complicated query tasks OLAP system needs high performance server, so that the cost of OLAP system is much higher for most middle-size and small-size enterprises. With the aid of distribution technology, we can deploy OLAP system on multiple minicomputers according to allocating tasks to ervery server, in this case can we attain application of OLAP with existing resource of enterprise. By using election algorithm in a distributed system the system coordinator is selected. In a full connected LAN, the adopted election schemas are generally the classic Bully algorithm by Garcia-Molina or improved ones. In this study, we propose a improved election algorithm which matches the distributed OLAP system that is more efficient with less messages.
出处
《河北软件职业技术学院学报》
2008年第3期52-55,共4页
Journal of Hebei Software Institute
关键词
OLAP
分布式
选举算法
欺负算法
OLAP
distribution
election algorithm
bully algorithm
