Assessing the Energy Impact of Java Software Debloating Tools

Abstract

Context: Software debloating tools are commonly used to reduce application size, minimize attack surface, and enhance performance, but may also offer potential for lowering energy consumption in Java applications.

Objectives: This study examines the energy-related impact of four Java debloating tools—ProGuard, DepClean, DepTrim, and JLink—by quantifying their effects on energy consumption and identifying factors that influence the energy use of debloated Java applications.

Methods: An empirical evaluation was conducted on 10 benchmark Java projects and 5 real-world Java applications. For each system, CPU and memory usage, execution time, power draw, and total energy consumption were measured across 30 independent trials, comparing debloated versions produced by each tool against the original baseline.

Results: Across all systems, energy and performance metrics remained largely consistent between debloated and baseline versions. While tool effectiveness varied by project, DepTrim achieved the greatest mean energy reduction, with a decrease of 0.80% in real-world applications.

Conclusion: The evaluated Java debloating tools yielded minimal improvements in energy efficiency and impact on the performance metrics for the tested systems, despite employing varied optimization strategies. These findings highlight the need for more energy-conscious debloating techniques and robust evaluation frameworks to promote sustainable software engineering practices. The methodology presented in this study can serve as a foundation for future work at the intersection of energy consumption measurement and software debloating strategies.