On heterogeneous distributed geoscientific query processing

Geoscience studies produce data from various observations, experiments, and simulations at an enormous rate. With proliferation of geographic applications, scientific data formats, and storage systems, interoperability remains an important geoscientific data management issue that is often overlooked in current geoscientific query processing research. We present how some issues concerning interoperability in geoscientific query processing are addressed in the Conquest parallel geoscientific query processing system under development at UCLA. The design of Conquest is based on the Volcano extensive query processing system which encapsulates parallel computation through the exchange operator. Conquest extends the exchange operator to support multicasting of data streams and hides the heterogeneity in hardware and operating system platforms from users developing parallel geoscientific applications. The Conquest data model captures some important structural and semantic properties of common geoscientific datasets and is used as a canonical model for a wide variety of scientific and non-scientific datasets. In addition, Conquest supports a uniform interface to a variety of scientific data sources. Access to data managed by a remote data repository is optimized by “pushing” Conquest operators into data repositories to maximize use of local database capability and reduce the volume of data transfer between the data repositories and Conquest