The need and opportunity to efficiently find patterns and features in the vast and growing scientific data sets of today is apparent. In this paper, we present an extensible and distributed query processing system called Conquest (CONcurrent QUEries over Space and Time) which delivers high performance for non-traditional database query applications. Conquest is composed of three components: query management subsystem, query execution engine and user-interface subsystem. Both operation and data parallelism are supported for high-performance query execution. Scientists can arbitrarily introduce their familiar well-known raster and vector geometry data types to the Conquest data model as well as particular algorithms - so-called operators - for specific applications. Tools are available to help support these new data types and operators so that a developer can focus on programming his/her core logic of algorithms without paying a penalty to deal with the novel execution environment’s features.