E-SECO ProVersion: uma arquitetura para manutenção e evolução de workflows científicos

Abstract

A scientific software ecosystem, in addition to other features, seeks to integrate all stages of an experiment, and commonly used scientific workflows to solve complex problems. Any changes that occurred in an experiment must be propagated to the associated workflows, which must be maintained and evolved for further successful research. One of the way to ensure this control is through configuration management. So that it can be used, it is important the storage of performance data and modeling of the experiment and associated workflows. In this study, we use the concepts and models related to the source of data for storage and retrieval of this data. Use the data source brings some advantages in storage and query, as we will see in this dissertation. Thus, this paper proposes an architecture to manage the development and maintenance of scientific experiments and workflows, called E-SECO ProVersion. The motivation for the specification and implementation of architecture came from the realization of a systematic review and a study to check maintenance characteristics and evolution in existing workflows repositories. From these analyzes, the main features of the architecture are defined and detailed. In addition, a roadmap with usage guidelines and proofs of concept using workflows extracted from myExperiment repository were presented in order to evaluate the applicability of architecture.