Design And Implementation Of A Data Integration Model Based On DDS And XML

The current enterprise computing environment is asynchronous, dynamic and loosely-coupled. Traditional data integration technologies can not meet the demand of distributing and getting updated data effectively. Publish/subscribe communication paradigm supports dynamic, asynchronous, many-to-many communication among loosely-coupled systems. It enables decoupling in space, time and control flow between communication participants. The Data Distribution Services for real time systems (DDS) Specification established by OMG can provide scalable, flexible, effective and real-time communication environment. In addition, it supports powerful application layer QoS control mechanism. However, most of current publish/subscribe middleware only supports topic-based matching. On the other hand, software implementations and performance tests of content-based matching have not been researched in-depth and thoroughly.After investigating data integration technologies and information distribution middleware technologies, this thesis designs and implements a data integration model based on DDS and XML which is suitable for the distributed heterogeneous communication environment. This data integration model implements the data-centric publish/subscribe communication paradigm proposed by DDS, and is flexible, loosely-coupled and scalable. The main innovations of this thesis include: 1. the model implements a subset of QoS policies proposed by DDS so as to provide powerful QoS control mechanism for data integration. 2. The model uses XML as the format of messages, Xpath to express subscriptions; by using the message agents—InfoRepo, it implements asynchronous, dynamic, scalable and loosely-coupled communication between data sources and matches the content of XML messages and XPath subscriptions in order to enrich the expressing ability of the system; messages are translated and transformed in the adapters. 3. An information distribution platform based on the data integration model is implemented and tested.The testing result shows: The data integration model based on DDS and XML proposed in this thesis implements content-based matches. It has good performances and can enable complete decoupling in time, space and control flow between publishers and subscribers. It provides technical supports for building enterprise data integration platform and facilitates the development of enterprise application systems.