Deliverable 4.1 State of the Art in RI and digital infrastructure sustainability and technologies assessment for energy efficiency and impact

The GreenDIGIT project is a European initiative aimed at reducing the environmental footprint of Research Infrastructures (RIs) and digital services by enhancing energy efficiency and sustainability. As digital infrastructures continue to grow in scale and complexity, their energy consumption and carbon emissions have become pressing concerns. This report, Deliverable D4.1, provides a comprehensive State of the Art assessment of current technologies, methodologies, and policies that influence the sustainability of RIs, with a particular focus on metrics, optimization strategies, and software solutions for energy-efficient operations.

The presented State of the Art (SotA) analysis provides a comprehensive overview of existing technologies, methodologies, and best practices for improving energy efficiency and enhancing the sustainability of scientific computing and digital RIs. It serves as a foundational study for WP4, WP5, and WP6, supporting the design, development, and deployment of GreenDIGIT’s sustainability-oriented frameworks and tools addressing aspects such as energy efficiency, carbon-aware computing, sustainable data management, and research reproducibility.

The deliverable assesses technological approaches and best practices across multiple infrastructure layers and the data collection and processing continuum, including cloud computing, high-throughput computing (HTC), 5G networks, and IoT environments, providing insights into how digital RIs can optimize their operations while minimizing energy consumption and carbon emissions.

The document is structured around major technology domains that play a crucial role in reducing the environmental impact of digital infrastructures: metrics and monitoring, scientific workflow scheduling and optimisation, energy efficient networks, research data management, experimental research reproducibility, and overall sustainable software design practices. Each chapter provides an overview of existing technologies, ongoing research, and potential improvements that will shape the next-generation sustainable RIs.

The deliverable provides an extended analysis of existing practices for data centre energy efficiency and carbon footprint metrics collection and monitoring. The report outlines the European Energy Efficiency Directive (EED) reporting requirements for data centres, highlighting metrics such as Power Usage Effectiveness (PUE), renewable energy usage, and water efficiency. Existing monitoring infrastructures like EGI, Euroean Open Science Cloud (EOSC), and Prometheus are assessed for their applicability in federated research environments. The study identifies gaps in carbon emission tracking and energy efficiency reporting, proposing an energy data model for harmonized monitoring across different RIs.

A key finding of this study is that RIs often lack the necessary tools and frameworks to effectively measure, monitor, and optimize their environmental impact. While significant advances have been made in energy-efficient computing, carbon-aware workload scheduling, and optimized data storage, their adoption remains fragmented. Scientific workflows are often executed without consideration of energy efficiency or carbon intensity, leading to excessive energy consumption and inefficiencies across HTC, high-performance computing (HPC), and cloud environments. Similarly, scientific data management strategies are not yet optimized for sustainability, with redundant data storage and inefficient replication practices contributing to unnecessary energy usage.

The presented analysis identified a need for GreenDIGIT to work on the development of federated environmental monitoring infrastructures, enabling RIs to track, report, and optimize their sustainability metrics in real-time. The project is also exploring energy-aware scheduling mechanisms, which will help align computational workloads with low-carbon energy availability. These advancements will significantly improve energy efficiency and carbon footprint management for scientific computing environments.

The overview and analysis of sustainable software design principles and practices will help the project to implement these practices in scientific applications design and efficient use. By enhancing Virtual Research Environments (VREs) and integrating intelligent experiment reproducibility frameworks, the project ensures that research workflows and experimental data collection can be reused and optimized, reducing redundant computations and minimizing energy waste.

Based on the presented State of the Art analysis, the deliverable provides key recommendations for energy efficiency and environmental sustainability of the future RIs:

• Standardized environmental metrics reporting across all European RIs, leveraging federated monitoring platforms.
• Adoption of energy-aware workload scheduling, integrating real-time carbon intensity data for optimized task execution.
• Implementation of sustainable software design practices, including energy-efficient programming, hardware-aware optimization, and carbon-aware software execution.
• Enhancing scientific experiment reproducibility through low-impact digital experiment frameworks, reducing redundant computational efforts and improving efficiency of research data management, in particular compliance with the Findable, Accessible, Interoperable, and Reusable (FAIR) data principles.

The report also touches on the need to address energy efficiency of the growing use of Generative AI (GenAI) in future scientific research.
The presented analysis and identified gaps and research and development topics will provide a basis for GreenDIGIT to develop and validate prototype necessary solutions and services, helping European RIs transition to greener, more energy-efficient operations. These efforts will contribute to reducing the environmental impact of scientific computing while maintaining high-performance research capabilities.

[This deliverable is pending approval from the European Commission]