eScienceSWaT -- Towards an eScience Software Engineering Methodology

Abstract

eScience is the field trying to provide IT support to scientists throughout the life cycle of scientific experiments. The major focus is on shortening the time to new discoveries and revealing knowledge about natural phenomena by providing software systems for different scientific tasks and for many domains. In this field, however, one can observe that there are no standardized and systematic approaches for developing the systems dealing with the processing and analysis of the large amounts of scientific data. In this paper we introduce eScienceSWaT, a software engineering methodology that aims at providing a systematic and user-friendly approach for building software applications for scientific experiments. The proposed approach considers the existing knowledge from computer science and software engineering but does not require scientists to be specialists in these fields. Furthermore, we present an architecture of a compos able system enabling eScienceSWaT. The system is meant to guide scientists in developing their experiments and features decision support capabilities based on patterns. Both the methodology and the architecture apply approaches from Enterprise Computing, Cloud Computing and Software Engineering. In order to demonstrate the suitability of eScienceSWaT, we apply it to a scenario from the field of scientific simulations.