Developing an architecture to support the implementation and development of scientific computing applications

The increasing complexity and computational demands of scientific computing applications has led to a need for robust software architectures to facilitate the conceptualization, design, implementation, deployment and maintenance of these applications. This paper attempts to shed light on how the unique characteristics of scientific computing applications, as well as computational scientists, make impositions upon the framework used to support scientific research efforts. We use our experience with NetSolve, a toolkit designed just for such interactions, as a means to present the approach of one infrastructure to support scientific computing. We further discuss how NetSolve implements the unique model of using a single system to aggregate, manage and access distributed hardware and software resources.