Software ecodesign: from requirements to operation

Abstract

The paper examines the concept of software eco-design as a crucial component of modern software engineering. It argues that the growing scale of digital systems, cloud computing, and artificial intelligence creates new challenges related to energy footprint and greenhouse gas emissions. Even minor optimizations in software solutions can yield significant global effects. The authors emphasize the importance of shifting from declarative approaches to measurable metrics such as carbon intensity, energy consumption, and data transfer volume. The study aims to establish a comprehensive eco-design framework that encompasses all stages of the software lifecycle, from requirements specification and architectural decisions to implementation, testing, deployment, and operational monitoring. Particular attention is devoted to two aspects: rational sustainability measurement through transparent metrics, and managed trade-offs between performance, usability, and environmental efficiency. The paper systematizes key practices: incorporating sustainability into non-functional requirements; architectural strategies for reducing data transfers and employing edge computing; algorithm and code optimization for resource efficiency; frontend approaches to minimize heavy content and third-party scripts; integration of energy profiling into CI/CD processes; and operational techniques such as autoscaling, night-time shutdowns, and carbon-aware scheduling. A special focus is given to data management and artificial intelligence, where efficient storage policies, data compression, model optimization, and balancing accuracy with computational cost are critical. Practical analysis demonstrates that the most significant impact is achieved at early stages, when sustainability is embedded into requirements and architecture. At the same time, the authors emphasize the importance of avoiding a dogmatic approach, ensuring that trade-offs remain transparent and evidence-based. The conclusions outline several perspectives for further research: advancing assessment methods so that energy and carbon metrics become as common as performance and reliability; developing reference architectural patterns and trade-off guidelines for various domains, from media platforms to financial technologies and AI systems; and integrating eco-design principles into software engineering education, ensuring that future professionals perceive sustainability as a fundamental engineering competence rather than an isolated initiative