I would agree that initially OGSA-DAI was essentially WS-JDBC plus some security. It rapidly extended to cover other data, such as XML, semi-structured files and other data forms. At present, several sites are adding RDF and SPARQL facilities. If it is just used in this way many of its benefits are missed. For example, the opportunity to compose data requests in one message round trip and the opportunity to compose activities so that redundant data movement is avoided. We have some writing on the design of systems using OGSA-DAI, but we need to expend more effort on that, in order that the potential benefits are understood by applications designers.
Projects extending OGSA-DAI bring in to focus the first major difference between OGSA-DAI and WS-JDBC: OGSA-DAI has three popular extensibility points, the data resource adapters, the activities and the client libraries. Each is extended by many research groups to develop tailored DAI functions. A good example is our colleagues in the GridMiner project led by Peter Brezney in Vienna. Another good example is the extension to OGSA-DAI for web data and RDF by the AIST group.
We have focused considerable effort on supporting extensibility. Partially because there is such a rich variety of data that it is essential, partially because this encourages further DAI R&D and partially because it provides a mechanism for bringing computation closer to data sources.
The next major difference is that OGSA-DAI contains a variety of multiple-data source functions, such as DQP (still in prototype), and multi-site query facilities which deal with partial availability.
There are also data transformation activities. These are being substantially extended in the forthcoming release. It is intended that they will provide a rich repertoire of composable tools to form powerful data-integration pipelined workflows.
It is true that it is highly desirable to raise the universe of discourse with data resources so that much of the detail of data schemas is hidden. We are very happy to work with any group that wishes to develop such high-level tools. We have to focus on pressing user demands and these have often been to gain access to the full facilities provided by the DBMS hosting the data. We are interested in tools that automate aspects of data access and transformation, but these need a consistent and robust infrastructure on which to work. We believe that we are developing OGSA-DAI so that it becomes a primary candidate for that infrastructure.