数据流上多滑动窗口聚集查询的优化算法  

An Optimized Algorithm for Multiple Sliding Window Aggregate Queries over Data Stream

作　　者：周心林[1] 赵雷[1] 

机构地区：[1]苏州大学计算机科学与技术学院,江苏苏州215006

出　　处：《小型微型计算机系统》2013年第4期774-777,共4页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(61073061)资助

摘　　要：在数据流的查询处理中,聚集查询是一种常用的查询类型.系统经常需要在同一个数据源上处理多个聚集查询,而单独地执行每个查询会导致严重的可扩展性问题和性能问题,因此实现相似查询之间的资源共享变得至关重要.针对多个具有不同时间窗口的聚集查询,本文提出了一种优化的窗口聚集算法OPWA(Optimized Paired Window Aggregation).先根据各时间窗口参数对聚集查询进行分组,使得相似查询可以同步调度;再采用paired技术对数据流进行分割.一方面减少了时间切片的数目,降低了空间的需求;另一方面同步地执行相似查询,减少了系统的计算开销.实验表明OPWA具有较好的性能.The aggregate query is widely used in query processing over data stream. Data stream systems often have to process multi- ple aggregate queries, since executing each query separately can lead to significant scalability and performance problems, it is extremely important to share resources of these similar queries by computing the common tasks only once. This paper presents Optimized Paired Window Aggregation for aggregate queries with varying time windows. According to the parameters of time windows, queries can be classified into different groups, all the queries in the same group are scheduled at the same time, and then paired windows chop a stream into unequal slices. This approach reduces the required buffer size by having relatively small number of slices and reduces the computation cost by re-evaluating the similar queries. Experiments show that this algorithm has good performance.

关 键 词：数据流 滑动窗口 多聚集查询 查询优化 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]