Carbon-Aware Routing Reduces Cloud Emissions Without Sacrificing Performance.

Modern cloud applications are often distributed across multiple geographic regions, but current routing mechanisms primarily focus on minimizing latency, often overlooking the environmental impact. This latency-first approach can lead to higher carbon emissions, especially when cleaner energy regions are available within acceptable performance limits. A new model addresses this by formulating the carbon-aware serverless routing problem as a constrained optimization task across geo-distributed cloud regions. It introduces an SLA-constrained carbon-aware routing policy designed to achieve optimal carbon reduction without compromising service level agreements (SLAs). Evaluated using real carbon intensity data from five primary AWS deployments, the proposed policy demonstrated remarkable results. It achieved up to a 46.8% reduction in carbon emissions while maintaining zero SLA violations across all tested thresholds. Under mixed workloads, the system reduced carbon by an average of 27.4%, with minimal routing overhead. A scalability study across 12 AWS regions further showed that average carbon savings could increase to 47.5% with greater routing flexibility. This work significantly contributes to climate action and clean energy goals by enabling low-carbon routing decisions in cloud systems.