Carbon-Intelligent Content Scheduling in CDNs
Sawsan El-Zahr, William Nathan, Noa Zilberman · 2025
Content delivery networks scaling with users' demand can meet quality of service needs by placing caches close to users. However, distributing content to caches is not for free. Not only it has a monetary cost, but also an environmental footprint. This work proposes a carbon-intelligent content delivery algorithm that accounts for cost and traffic demands as well as carbon emissions. Given the spatial and temporal dynamism of carbon intensity, this work aligns data delivery to caches with low carbon intensity periods and low distribution costs. The solution is evaluated on the Netflix Open Connect network, showing operational carbon savings between 13% and 64% while conserving other constraints.