Evaluation of Two XML Storage Approaches for Scientific Metadata

Abstract

Scientific data are increasingly described by metadata based on detailed XML schemata that capture both general and domain-specific concepts about the underlying data. Metadata captured using detailed XML schemata tailored to specific scientific domains increases the potential for data reuse by providing the ability to discover data products described by detailed concepts. Since such metadata is captured as XML, one alternative for managing scientific metadata is to store and query the metadata using a native XML database. Our research shows that a hybrid XML-Relational structure such as is used in the XMC Cat metadata catalog outperforms a native XML database for storing and querying scientific metadata; and significantly outperforms the native XML database under a scaled workload of concurrent inserts and queries.