【摘要】 并行查询中的处理机分配是并行数据库系统中查询优化的重要组成部分.考虑到进程对CPU和磁盘IO的需求,着重研究SN结构上并行查询的处理机分配问题.首先提出查询操作对CPU和IO的二维资源使用模型,再通过粗粒度概念确定(浮动)操作的划分并行度,并定量分析查询内各种并行性和资源共享对查询响应时间的影响,最后提出HLS启发式处理机分配算法更多还原

【Abstract】 Taking account of operators needs of CPU and disk I/O into account, we focus our research on processor allocation of parallel query on Shared Nothing execution environment. A resource usage model of two important resources (CPU and Disk I/O) is presented, and the degree of partitioned parallelism is derived on the basis of the quantification of the notion of coarse granularity. Using this model, we analyze the effects of three forms of intra query parallelism and resource sharing on query response time. An HLS algorithm for scheduling independent query tasks (pipelines) is presented.更多还原