The volume of digital scientific data has increased considerably with advancing technologies of computing devices and scientific instruments. We are exploring the use of emerging Grid technologies for the management and manipulation of very large distributed scientific datasets. Taking as an example a terabyte-size scientific database with complex database schema, this paper focuses on the potential of a well-known Grid middleware—OGSA-DQP—for distributing such datasets. In particular, we investigate and extend the data type support in this system to handle a complex schema of a real scientific database—the Sloan Digital Sky Survey database.