The Relational Grid Resource project explores the representation of grid resource data. Grid information includes information and attributes about resources in a computational or data grid, including compute servers, clusters, storage resources, people, groups of people, etc. We posit that the needs of the computations that run on a grid impose a unique set of demands on a resource repository that are not being met by the current hierarchical or attribute-value approaches to directory services. Specifically, user-driven needs necessitate support for complex queries over the resource repository and the timeliness needs require policies about update rates to the data contained in the repository. Our approach explores relational database-based technology for management of grid resource information.
This work is being done in conjuction with the Unified Relational Grid Information Services project under Peter Dinda at Northwestern.
The benchmark/workload is a set of queries and 'scenarios', or short-duration workloads, developed from a common representation for resource data that is neutral to a particular data model.
We apply the benchmark to three platforms that could viably host a grid information server: one platform is a relational data model and uses SQL as the query language. The second and third are hierarchical data models: an XML system that uses XPath, and MDS2 which uses LDAP as its query language. Included also are scripts that operate on a relational database through the UK National e-Science Institue's OGSA-DAI.
Beth Plale, Craig Jacobs, Scott Jenson, Ying Liu, Charlie Moad, Rupali Parab, and Prajakta Vaidya, Understanding Grid Resource Information Management through a Synthetic Database Benchmark/Workload conference submission, under review, November 2003.
Beth Plale, Craig Jacobs, Ying Liu, Charlie Moad, Rupali Parab, and Prajakta Vaidya, Benchmark Details of Synthetic Database Benchmark/Workload for Grid Resource Information Technical Report TR-583, August 2003.
Beth Plale, Peter Dinda, and Gregor von Laszewski,
Key Concepts and Services of a Grid Information Service,
ISCA 15th International Parallel and Distributed Computing
Systems (PDCS), September 2002,