Using XML and XQuery for data management in HPSS

The XML and XQuery language capabilities of modern databases can provide a powerful and flexible method of data management within a mass storage system. Within the High Performance Storage System (HPSS), the implementation of XML and XQuery capabilities for user metadata is called “User-defined Attributes” or UDA. The UDA feature provides a method for user applications to associate arbitrary metadata with HPSS namespace objects and store it in an organized, scalable, and searchable manner using XML. The implementation includes a simple key-value interface as well as exposure of the database´s XQuery interface to allow for highly customized and atomic update, retrieval, and namespace-wide search requests. Using this architecture enables HPSS to provide client applications a high degree of flexibility in the storage, management, and access of user-defined metadata. This paper describes a brief history of data management within HPSS as well as the architectural decisions, implementation, and results of the UDA feature. Also discussed are considerations for planning and management of the UDA feature, current and in development UDA solutions created by developers and customers, and possible future data management work within HPSS.