Query Process And Optimization For Active Real-time Embedded Database

Real-time database systems (RTDBS) need to face the challenge of simultaneously satisfying data integrity and timing requirements. The transactions and data in real-time database systems contain explicit time constraints, the correct criterion of transaction execution must satisfy both timing and logical constraints. Active rule systems are seamlessly integrated into real-time database systems for the reactive models. The high performance of RTDBS needs the main-memory database system (MMDB)'s support, and MMDB insures that the transaction has no I/O access in its running time. According to the above features, it is an important domain to improve the traditional query process and optimization mechanism and introduce it into embedded real-time database management system (ARTs-EDB).Based on ARTs-EDB, which is an active real-time database system we ourselves developed, our research discuss with emphasis on the design and implementation of query process and optimization mechanism embodying the MMDB's features.With respect to the characters of ARTs-EDB and referred to traditional query process mechanism of relational database system, the architecture of the query process mechanism in ARTs-EDB are designed and it contains the main modules in query process and interfaces to other subsystems.Query execution is the basis of query process mechanism. The adaptive capacity and executing performance of query execution will work on the whole performance of ARTs-EDB to some extent. Considering to improve them, we discuss an abstract query executing model based on some popular executing modes in detail. And we advance the pipeline executing mode by combining the multi-thread technique, which greatly improve the executing performance.Query optimization is very important to improve the performance of query process. We need consider many factors while designing the optimization module. Among them, the cost formula of plan tree and search strategy on searching space of plan trees are crucial. According to the MMDB's features, the join tree and costly factors are analyzed, and then we get the dynamic programming searching algorithm based on the specified join tree type and the cost formula which adapt to ARTs-EDB.