Scoped and approximate queries in a relational grid information service

We are developing a grid information service, RGIS, that isbased on the relational data model. RGIS supports complexqueries written in SQL that search for compositions (using joins)of resources. For example, we might ask it to find a Linux clusterwith a certain bisection bandwidth and total memory. Such queriescan be expensive to execute, however, and so we have developedseveral approaches that leverage our GIS schema to let us tradeoff between the number of results returned and the execution time.In this paper, we describe two of them: scoped queries and approximatequeries. Scoped queries constrain search to a networkneighborhood, returning all matching results in the neighborhood.Approximate queries reduce the number of joins done by replacingcollective constraints with constraints on individual resources,returning a subset of all the possible results in the grid. Scoping,approximation, and nondeterminism (described elsewhere),can be combined. In this paper, we describe scoped and approximatequeries, how they are implemented, and present performanceevaluations for two examples. The evaluation suggests that scopingand approximation can greatly reduce query times while stillreturning a useful number of results.