多事件模式并行CEP处理研究

Research on Parallel CEP Processing with the Multi-Event Pattern Sharing Capability

为弥补复杂事件处理技术(Complex Event Processing,CEP)在大规模事件并行处理、共享事件关联以及整体运行能力优化等方面存在的不足.文中给出了一种进行了时间扩展的有色Petri网检测模型.利用该模型将CEP事件模式原子单元形式化描述为基本Petri网结构,同时给出了网结构的三种连接策略,以及多事件模式的合并算法.研究结果表明:以文中检测模型为基础的原型系统,能够有效共享中间结果,并且利用并行执行的进程有效减少总体处理时间.当同时运行的事件模式数量越大时,系统运行效率越高.

To overcome the deficiencies of CEP （complex event processing ,CEP） technology in the aspects of parallel streams processing ,the correlative event pattern sharing and the overall efficiency optimization ,the paper presents a new colored petri-net detection model with some temporal extension . Based on the model ,the atomic units of the CEP event pattern are formalized into some basic petri-net structures .Three connection strategies for constructing event pattern and the multi-event pattern merge algorithm are also given . It can be concluded that the system ,based on this proposed model ,can effectively share the intermediate results ,and reduce the overall processing time by executing more threads .With the increasing number of running patterns ,the efficiency of the system becomes higher .