CORBA and XML: Design Choices for Database Federations

The newly established standards of CORBA and XML make it much easier to interoperate between different database software running on different platforms. We are using these in a mediator-based architecture that supports integrated access to biological databases. We discuss, in turn, design issues that arise from using each of the standards. In CORBA an important design issue is the use of coarse grain access, which supports a query language over an extensible integrated data model, as compared with fine grain access, which is tailored for specific queries. We discuss experience in using CORBA in these two ways. On the other hand we describe scenarios where returning results are communicated in XML format. We present a classification based on design choices.