Managing data in multiple, heterogeneous databases across an enterprise is a difficult problem. Heterogeneous database management systems (HDBMSs) begin to solve this problem by providing ways to specify what information should be accessed in the existing databases. However, these systems typically do not provide any way to declare how the integrity of the global or federated structure, and the data which populates it, should be maintained during global update operations. Instead, users must rely on their own knowledge of the distribution of the data and the relationships between the back-end stores to write application programs that maintain these relationships....[ Read more ]

Managing data in multiple, heterogeneous databases across an enterprise is a difficult problem. Heterogeneous database management systems (HDBMSs) begin to solve this problem by providing ways to specify what information should be accessed in the existing databases. However, these systems typically do not provide any way to declare how the integrity of the global or federated structure, and the data which populates it, should be maintained during global update operations. Instead, users must rely on their own knowledge of the distribution of the data and the relationships between the back-end stores to write application programs that maintain these relationships.

The objective of this research is to define a language that allows users to declare how to maintain the integrity of a global data structure, the parts of which are distributed across heterogeneous databases. At run-time, a query translator uses the specification to execute correct updates and queries to the back-end systems. Users do not have to create special views to represent the distribution scheme or the interdatabase integrity constraints as part of the application. Furthermore, the language modularizes interdatabase dependencies and relationships from global schema definitions so that each can be maintained and evolved separately. This design will allow different user groups to crea.te views of the information that can be maintained separately from the integrity constraints across the local databases.

A prototype is implemented to demonstrate the core concepts of the language. A design has also been included to show how it could interface with an existing heterogeneous database management system.