Interoperability of Multiple Autonomous Databases .

Abstract

Database systems were a solution to the problem of shared access to heterogeneous files created by multiple autonomous applications. To make the data usage easier, one proposed to replace the autonomous files by a globally integrated collection of data called a database. The idea was successful to a large extent and there are now many databases distributed over local and longhaul networks and frequently even on the same larger computer. Unavoidably, users now need shared access to multiple autonomous databases. The question arose as to what the corresponding principles should be. Should one reapply the database approach principles one level up or should new methodologies be introduced? We show that new methodologies have appeared, defined specifically for the management of multiple autonomous databases. They lead to a new type of systems, called multidatabase systems or federated systems. These systems make databases interoperable, ie manipulable together in a non- procedural way, without global integration. They also preserve the autonomy of each database to satisfy first its own needs. Systems of that type will be of basic importance, especially for distributed databases and we analyze the corresponding reasons. We also present the methodologies proposed for their design and discuss their relationship. We further show that the evolution towards multidatabase systems is already on the way, as mapr commercial relational database systems are becoming of this type. We discuss their capabilities and limitations with respect to advanced prototypes. We also show industrial prototypes and the standardization issues. Finally, we present some research issues.