Design And Development Of Distributed Service Framework For Real-time Database In Process Industrv

With the development of process industry, the amount of data generated in a variety of industrial automation systems grows rapidly. Due to the limitation of its capacity and performance, traditional stand-alone real-time database cannot satisfy the need of massive data processing in large-scale applications. Distributed service framework for process industry can disperse the amount of processing that a machine cannot afford to multiple machines, resulting in better scalability. But the related study is insufficient. Therefore, the study of distributed service framework for real-time database to improve system capacity, availability and real-time performance is of great significance and value.By studying the existing real-time databases and distributed technologies, and through analysis of functionality and performance requirements for distributed service framework in process industry, we proposed an easily extensible distributed service framework suitable for massive data processing based on existing service bus, supporting integrated management of heterogeneous services. System capacity is improved through multiple stand-alone real-time databases access as a whole. Master-slave/multiple redundancy schemes and concurrent access mechanism were designed to improve system availability and access performance.In this thesis, the overall structure of the distributed service framework was designed using the idea of hierarchy. The main work is as follows:(1) Service registry was designed to solve the Id assignment, calibration and other management issues. Then, intelligent path finding strategy was designed based on Dijkstra’s algorithm to quickly obtain the optimal path between two points. One-stop service operation and monitoring is achieve through automatic deployment mechanism.(2) Distributed query and subscription services were designed to manage the lifecycle of client requests. The complexity of the stand-alone real-time database is masked through real-time database proxy, providing unified access interfaces. Stand-alone real-time databases can quickly access the framework through these services without changes, which breaks the performance bottleneck of single computer, and improves system capacity.(3) For high availability requirement, the real-time database and central services master-slave redundancy schemes were designed; for other services with no center, multiple redundancy scheme was designed, thus the system availability is improved, solving the problem of single point of failure. For real-time requirement, distributed concurrent access mechanism was designed using the thread pool and task queue technology. Task fragmentation and task priority was introduced to respond to large number of requests in time, improving the efficiency of data access.Based on the above design, a distributed service framework prototype software was implemented and a test platform was built using ordinary commercial computers to test the functionality, performance and availability of distributed service framework. Test results show that distributed query and subscription of massive data are effectively achieved through the framework. Concurrent query response time meets the real-time requirement. As long as one instance of service works, the system can continue to provide service, ensuring high availability of the system.