Model-based software sustainability analysis and improvement

Abstract The environmental sustainability of software has become a significant issue for the use of computing technologies. The pervasive use of software applications throughout society has potential negative environmental impacts due to the energy used by software execution, and the resulting greenhouse gas (GHG) emissions. In this paper, we describe techniques for the identification of software energy use flaws at the software modelling level, and we define techniques for the removal of such flaws via refactoring or the use of alternative specification/design modelling elements. The key idea is that energy use flaws are corrected once at the software modelling level, in order to obtain energy-efficient benefits across all target platforms. We provide a detailed evaluation to show that the proposed improvements do result in reduced energy use across multiple implementation languages and platforms. The outcome represents a novel approach for software sustainability using software models, incorporated into an established toolset for model-driven engineering.