Service Infrastructure for Cross-Matching Distributed Datasets Using OGSA-DAI and TAP

One of the most powerful and important goals for VU developers has been to enable cross-match queries between disparate datasets for end users. This has only been achieved within the VU using the early SkyNode infrastructure and has not been reproduced using current IVOA standards. To remedy that situation, the Wide Field Astronomy Unit (WFAU) has worked with the Edinburgh Parallel Computing Center (EPCC) in leveraging the OGSA-DAI grid middleware to enable cross catalog queries on distributed VU services. We have achieved this goal by building a three layer service stack that places the OGSA-DAI software above multiple individual services implementing the IVOA's new Table Access Protocol (TAP), and then a single TAP service is placed above this and presented to the end users. Users can then execute ADQL queries that cross-match between the disparate datasets as though they were in the same database with acceptable performance rates on the resulting data flow. The OGSA-DAI software is able to interrogate any compliant TAP service to acquire the necessary metadata for insertion into the single federated TAP service used for cross-match queries. We are currently testing this distributed infrastructure using the TAP services provided by WFAU for the UKIDSS DR3 and SDSS DR7 datasets in combination with the TAP service available from the Canadian Astronomy Data Center (this last without requiring any action from CADC staff). This forms the basis for a large-scale distributed data mining workflow and similar activities can be readily implemented as more TAP services come online. Future work will involve releasing this infrastructure to the greater astronomical community as an IVOA compliant service for users.