一种基于优先级的数据流查询实时调度策略

Real-time Query Schedule Strategy Based on Priority in Data Stream System

针对一些高关键性的特殊应用,把传统实时系统中的截止期、空闲时间、关键性等概念引入数据流管理系统,根据系统中连续查询的特征,赋予其新的涵义。提出了一种基于优先级的实时调度策略,使得流截止期较早、空闲时间较短、关键性较高的查询具有更高的优先级;给出了一种优先级树结构,使得基于优先级的查询执行序列唯一化。实验表明该策略提高了连续查询的实现价值率和连续查询调度的成功率。

Focusing on some high critical application, the conceptions of deadline, slack and critica in real-time system are introduced and improved in DSMS according to the characteristics of continuous query in DSMS. According to these new conceptions, a real-time schedule strategy based on priority is proposed. In this scheduling strategy, the earlier the deadline is or the shorter the slack is or the more critical the query is, the higher the priority is. And a structure of priority tree is proposed to realize the unique execution sequence of priority-based query. The experimental results indicate that the strategy raises the hit value ratio （HVR） and the success ratio of continuous query scheduling.