Energy Consumption of Behavioral Software Design Patterns

Abstract: The environmental and economic implications of the increase in Information and Communication Technology energy consumption have become a topic of research in energy efficiency. Most studies focus on the energy estimation and optimization of lower tiers of the hardware and software infrastructures. However, the software itself is an indirect driver of energy consumption, therefore, its energy implications can be to some extent controlled by the software design. Software design patterns belong to high-level software abstractions that represent solutions to common design problems. Since patterns define the structure and behavior of software components, their application may come at energy efficiency costs that are not obvious to the software developers. The existing body of knowledge on energy consumption of software design patterns contains a number of gaps, some of which are addressed within the scope of this thesis project. More specifically, we conducted a series of experiments on the estimation of energy consumption of Visitor and Observer/Listener patterns within the context of non-trivial data parsing in Python. Furthermore, we considered a Patternless alternative for the same task. Additionally, our measurements include runtime duration and memory consumption. The results show that the Visitor pattern led to the largest energy consumption, followed by Observer/Listener and finally the Patternless version. We found a strong relationship between runtime duration and energy consumption, thus coming to the conclusion that the longest-running pattern is the most energy-consuming one. The findings of the current study can be beneficial for Python software developers interested in the energy implications of software design patterns.