Research And Implementation Of Heterogeneous Data Integration And Exchange System Based On Federations

With the development of information construction, many application subsystems have been established according to different kinds of needs in various organizational units. Because of the lack of unified planning about different business logic and data sources, these subsystems are regarded as "isolated information islands".Under the background of business connecting each other closely, it urgently needs to make these separate subsystems be an organic union and share information with each other. Therefore, an integrated system need to be established to unite various separate subsystems. In the integrated system, these subsystems are mutually independent and can easily definite the information, obtain others information, exchange data with each other and get the transparent access to global data. It is a worthy of study subject to implement such an integrated system.For the above reasons, the integration and exchange of data is deeply studied in this thesis, and a system solution which adopt the federal model is put forward. It can easily achieve the goal of automatic exchange between distributed data and get transparent access to the global data.The advantages and disadvantages of existing data integration methods are analyzed. In this thesis, the integration and exchange system of data is designed starting from the requirements. Through the system, users could define the global data,get the access to the data and extract them.Through this integrated system, various subsystems could build a unique virtual database by data definition, data publication and subscription and a series of operations.Besides, the control center in the system is responsible for system’s maintenance, control and management, as well as the publish-subscribe relationship in the data container. Each subsystem as a member of federation push or pull data according the data container, achieving the goal of sending sharing data automatically and artificial query of the needed data. When a subsystem need to update the data, it can send and receive data based on the publish/subscribe file, which is called push operation. Push operation is commonly used when the data is dynamic, and this kind of data is always real-time data or delayed data. In addition, the transparent access to the global data from the perspective of the user is realized in this thesis. Users could get any data structure and content through the query of virtual database. Three mode conversion among user-model, global-model and member-model achieve the purpose of the data pull operation. Pull operation is used for historical data.The data integration scheme adopts the extensible and flexible XML technology, the cross-platform of Java technology, and uses the SOCKET network communication technology as the technical support. In this system, each subsystem has high flexibility. For the good portability and scalability of the integrated system, it has a certain universality.