Interoperating with DIF data

Most computerized data is not under the purview of traditional database management systems, but, rather, it is stored in files. Data Interchange Formats (DIFs) are the basis of an important class of such files, containing massive volumes of data, and serving a large user community. Initially designed for data exchange, DIFs are increasingly being used for data storage. Collections of DIF files have become de facto databases, and the software tools surrounding them have evolved to become ad hoc data management systems. But these systems are primitive compared with traditional database management technology. They lack flexibility, and are unlikely to scale as DIF data volumes grow. An acute need exists to provide more robust DBMS style support for data in such files, and to support interoperability in heterogeneous environments that include this data. We suggest that object oriented database techniques can be used to provide effective data management support for individual DIFs and provide a platform for interoperating with heterogeneous DIF data, both in a scalable fashion. We present architectures for implementing this support, and outline the research issues they engender. We present initial results of an experimental evaluation of one such approach.