Nested Pattern Queries Processing Optimization over Multi-dimensional Event Streams

Recently, many modern applications have required complex event processing technology to analyze multi-dimensional stream big data available in real-time data feeds. To address the above requirement, we develop a novel real-time event stream processing method called multi-query optimization strategy (MQOS). MQOS aims at exploiting not only common sub-expressions among nested event pattern queries, but also replicas of the appropriate common operators´ results for the queries needed to minimize recalculation and re-communication costs. We first design a triaxial hierarchy consisting of nested query pattern, nested query concept and operator type hierarchies to specify the relationship among the sub-expressions of queries. Next, based on the triaxial hierarchy, we devise a cost-based heuristic to find an optimized query execution plan with minimum costs of operators and communications. We then propose three reuse schemes of common sub-expressions: nested query pattern-based, nested query concept-based and operator type-based reuse schemes. By integrating the optimized query execution plan-find approach with the three reuse schemes, we present the MQOS to achieve nested pattern queries processing optimization. Finally, our experiments tested on StreamBase under different workload conditions demonstrate the superiority of MQOS.