Enterprise information integration – XML to the rescue!

Abstract

The database field has been struggling with the data integration problem since the early 1980’s. We’ve named and renamed the problem – heterogeneous distributed databases, multi-databases, federated databases, mediator systems, and now enterprise information integration systems – but we haven't actually solved the problem. Along the way, we've tried data model after data model – functional, relational, object-oriented, logical, semistructured, you name it, we've tried it – and query language after query language to go with them – but we still haven't solved the problem. A number of startups have died trying, and no major software vendor has managed to hit a home run in this area. What's going on? Is the problem too hard? Should we just declare it impossible and give up? In this talk, I'll explain why I believe now would be exactly the wrong time to give up. After a brief look at history, I'll make the case that we are finally on the verge of finding a real solution to this problem. I'll define the enterprise information integration problem as I see it and then explain how the XML and Web Services revolutions that are in progress – based on SOAP, WSDL, XML Schema, XQuery, and so on – relate to the problem and its solution. I'll describe the path that we are on at BEA to deliver a solution, and finally I'll leave the audience with my thoughts on some open problems where the database field, especially the “modeling crowd”, can contribute.