Datacenter Energy Consumption Optimisation by DNS EvaluationS – DECODES

The DECODES (Datacenter Energy Consumption Optimisation by DNS EvaluationS) project tackles the urgent challenge of rising energy consumption in data centers by focusing on a critical but often overlooked component: the Domain Name System (DNS). While DNS traffic accounts for only a small share of Internet data, its continuous operation and highly distributed architecture, spanning recursive resolvers, authoritative servers, root servers, and interconnection points such as IXPs, result in a significant cumulative energy impact. As global data centers and networks strive to become more sustainable, DNS represents a key area where optimisation is both possible and necessary.

To address this, DECODES introduces a comprehensive methodology to measure, model, benchmark, and optimise DNS energy consumption across diverse operational scenarios. The project emphasises the need to balance performance and security, particularly in light of new protocols such as DNS-over-HTTPS (DoH), DNS-over-TLS (DoT), DNSSEC, and the anticipated integration of Post-Quantum Cryptography (PQC) into DNS infrastructure. These emerging protocols, while enhancing security and privacy, introduce computational and transport-layer overhead that may further increase energy usage. Understanding and mitigating these trade-offs is a central objective of the project.

DECODES proposes developing a standardised measurement framework that enables fine-grained, interoperable monitoring of energy metrics across heterogeneous DNS systems. This framework builds upon emerging IETF standards. It will be implemented through a combination of hardware instrumentation (e.g., wattmeter’s), software telemetry, and passive and active network probes. The framework will be validated in real-world deployments and a controlled testbed that enables reproducible, scalable experimentation with realistic traffic and infrastructure configurations.

Beyond measurement, the project will generate reference energy models based on a multi-level analysis: macro-level (infrastructure-wide inventory and traffic profiling), micro-level (energy impact of query types, encryption, and caching mechanisms), and user-level (perspectives of end-users, resolver operators, and authoritative server administrators). These models will guide the design and evaluation of optimisation strategies, including TTL tuning, energy-aware resolver selection, query minimisation, efficient anycast routing, and adaptive link sleeping techniques.

DECODES is deeply committed to open science and reproducibility. All anonymised datasets, software tools, energy models, and test configurations will be published on open-access platforms. The project will also engage with the wider technical community through participation in IETF and RIPE hackathons, technical demonstrations, and knowledge-sharing sessions.

By collaborating with key actors in the DNS ecosystem in France: Afnic (registry and authoritative operator), France-IX (interconnection provider), alongside leading academic institutions (Télécom Sud Paris, École nationale des ponts et chaussées), DECODES ensures access to a broad, representative range of DNS infrastructures for validation and impact.

Furthermore, DECODES plans to contribute to standardisation efforts within IETF, ITU-T, and other international bodies, aiming to influence best practices and technical recommendations for energy-aware DNS operations. In summary, DECODES aspires to be the first major initiative to propose and validate a standardised, scientific approach to quantifying and optimising the energy footprint of DNS infrastructure, bridging the gap between security, sustainability, and operational efficiency.