复杂事件处理中多聚合查询共享方法

Multi-aggregate query sharing method in complex event processing

复杂事件处理技术是在持续不断的流数据中检测满足特定事件序列或对匹配的事件进行统计的一种流数据处理技术。在处理带有Kleene操作符的事件趋势聚合查询时,需缓存中间结果来实现不定数量事件序列的匹配,故对查询系统的资源需求较大。利用多个查询之间存在的共享机会,生成用于指导查询处理的共享计划,可以有效提高事件趋势聚合查询的处理效率。但现有的聚合查询处理方法生成的共享计划无法做到随执行环境的变化而进行实时调整,来支持查询系统的持续高效处理。针对上述问题,提出了一种可以动态更新的多聚合查询共享方法,以支持实时变化的复杂事件检测的持续高效处理。通过提出共享图数据结构和代价模型,实现对所生成共享计划的实时调整,并引入在线增量聚合执行共享方法,进一步提升带有Kleene操作符的事件趋势聚合查询的处理效率。在真实数据集和模拟数据集上分别进行实验,并与其他处理聚合查询的方法进行了实验对比。实验结果表明,提出方法能够有效降低查询延迟,提高整体查询的处理性能。

Complex event processing technology is a streaming data processing technique that detects specific event sequences or performs statistics on matching events in continuously flowing data.When processing trend aggregation queries with Kleene operators,caching intermediate results is necessary to match an indefinite number of event sequences,thus requiring significant resources from the query system.Exploiting opportunities for sharing among multiple queries,generating shared plans to guide query processing can effectively enhance the processing efficiency of trend aggregation queries.However,existing me-thods for handling aggregate queries do not dynamically adjust the generated shared plans in real-time to support continuous and efficient processing of complex event detection in response to changes in the execution environment.Addressing these issues,this paper proposed a dynamically updatable method for multiple aggregate query sharing to support the continuous and efficient processing of real-time changing complex event detection.By introducing the shared graph data structure and cost model,this method achieved real-time adjustment of the generated shared plans.And by using the online incremental aggregation execution sharing method,it further enhanced the processing efficiency of event trend aggregation queries with Kleene operators.This paper conducted experiments on both real and simulated datasets,compared the method with other approaches for handling aggregate queries.The results of the experiments indicate that the method effectively reduces query latency and improves the overall processing performance of queries.