Data Grids A data grid is a management system that organizes digital entities for use. These systems provide a logical name space for referencing digital entities, and provides services that map distributed information to the logical name space. These systems manage consistency requirements that define how the information from different services can be integrated. These capabilities provided by data grids are expected to simplify the implementation of archival processes, both in terms of decreasing the amount of time that is needed to apply the archival processes, and in terms of increasing the size of the collections that can be managed. Persistent Archives Reagan Moore is the organizer of the Persistent Archive Research Group within the Global Grid Forum. He is also a participant in this project. Storage Resource Broker Developed at SDSC, the Storage Resource Broker (SRB) technology has been employed in multiple data grid and persistent archives projects. These include prototypes for NARA and for the National Science Foundation's National Science Digital Library. In addition, the SRB is currently used to support digital libraries that are publishing collections, and it is used to support the sharing of large collections. The SRB system at SDSC currently supports over 7.6 million registered digital entities, comprising over 50 terabytes of data. The system has been used to move data at rates exceeding 20 Mbytes/second between SDSC and the National Center for Science Alliance in Illinois. These prior applications demonstrate that the technology is sufficiently robust to support the Persistent Archives Testbed. Archival Processes This project will test and evaluate the ability of the SRB data grid technology to partially automate the archival functions of appraisal, accessioning, arrangement, description, and access. Within the testbed, automation of at least one archival process will be explored at each site. This will involve additional capabilities of the SRB, ranging from extraction of descriptive metadata, to creation of sub-collections for a record series, to replication of containers.

February 16, 2004