Heterogeneous Data Integration Middleware Integration Engine Design And Realization

Data integration solves the problem of Information Island, but designing and implementing a large scale distributed data integration mid-ware is not simple. It meets some problems, such as integrating the legacy systems effectively, handling the massive concurrency and stability. This dissertation presents a well formed architecture, Service Oriented Event Driven Architecture (SOEDA), for data integration mid-ware engine. This architecture improves the cooperation ability, smartness, and integration ability of the application systems. It also supports massive concurrency data integration service and it is robust to huge variation in load.In SOEDA, the data integration application is regard as a network of "service nodes" which are connected by event queue. Service node is a software component which supplies the concrete data integration functions. The service nodes implement the standard service interface and register to the service center. The only thing you just do is to wrap resources of the enterprise into the service node, but need not to modify them, and then you can integrate all the resources of the enterprise smartly to form a data integration platform. Using of the event queue enables the controlling of the service node according to the load and resource condition of the data integration system.I present a performance model for the data integration engine implemented according to SOEDA. The engine is modeled as a load-aware open network of service nodes based on queue theory. The queue length and the response time of the engine are formally argued using the queue theory. It argues that the SOEDA is effective for the massive concurrency. The modeling involves Numerical Statistic, Math about Uncertainty, Formalization Theory. SOEDA follows the idea of the service oriented architecture and adopts lots of design patterns. Use many java techniques to implement the data integration mid-ware engine, such as JMX, JBI, and JMS etc.Our data integration mid-ware engine is modeled and designed based on masses of formal theories, and implemented accord to the related specifications and adopting the newest technical. So, I believe that it has a good market prospect.