Production-level distributed parametric study capabilities for the grid

Abstract

Though tools are available for creating and launching parameter studies in distributed environments, production-level users have shunned these tools for a variety of reasons. Ultimately, this is simply a result of the inability of these tools to provide anything more than a demonstration-level capability, rather than the flexibility and variety of industrial-strength capabilities that users actually require. In addition, despite the difficulties of creating parametric studies without specialized tools, users still demand that such tools be intuitive, easy to use, and versatile enough to support their particular experimental procedures. We show some solutions to real problems encountered in users’ parametric experiments, and simultaneously show how the success of grid computing in general will rely on the ability of grid tool developers to provide a much greater level of capability and generality than users have seen in current grid-tool demonstrations.