GreenOps Is FinOps With a Second Bill: Carbon-Aware Scheduling in Practice

FinOps, the practice of managing cloud costs, is expanding to include a second bill: carbon emissions. GreenOps is the practice of treating carbon costs as a real, measurable, and optimizable number, just like FinOps treats dollars. Most engineers should pay attention to GreenOps because most actions that reduce carbon also reduce costs.

Carbon-aware infrastructure has two primary levers: when you run your workloads and where you run them. Time-shifting is the first lever, referring to the fact that the carbon intensity of the electricity grid varies throughout the day. Running batch jobs during hours with cleaner energy, such as nighttime or midday with abundant solar power, can significantly cut carbon emissions. This concept is similar to running non-production workloads during off-peak hours to save costs but with a different focus on carbon.

The second lever is region-shifting, which involves running workloads in cloud regions with different energy mixes. Cloud regions powered by renewable energy sources, such as hydro or nuclear, produce significantly less carbon emissions per compute-hour compared to regions reliant on fossil fuels like coal. Workloads that are not latency-sensitive can be moved to greener regions, which offers a cost-saving and carbon-reduction opportunity.

The good news is that GreenOps often overlaps with existing FinOps efforts. Identifying and killing idle resources, rightsizing resources, and shifting non-production workloads to off-peak hours all contribute to both cost savings and lower carbon emissions. These actions share the same decision-making process, making it easier to implement GreenOps alongside FinOps initiatives.

However, there are some differences between the two. A cheaper cloud region may not always be a greener region, as some low-cost regions may run on dirtier grids. When optimizing for carbon, you may occasionally choose a slightly more expensive region for cleaner energy. This tradeoff requires careful consideration and decision-making to balance cost and carbon emissions effectively.

While carbon-aware scheduling can conflict with latency requirements, it is most beneficial for non-user-facing jobs that can tolerate delays. A progressive approach to implementing GreenOps involves measuring carbon emissions first, then optimizing by implementing cost-saving measures like killing idle resources, rightsizing resources, scheduling non-production workloads, and raising utilization. Reporting both cost and carbon savings alongside each other highlights the dual benefits of GreenOps initiatives.

The urgency to adopt GreenOps stems from two main reasons. First, carbon emissions reporting is becoming a regulatory and procurement requirement, making early measurement easier to comply with. Second, presenting cost optimization efforts as GreenOps initiatives can increase funding and priority from executives, as the combined impact of cost and carbon savings is a single, compelling project.

In summary, GreenOps is FinOps with a focus on carbon emissions. By understanding and leveraging the two key levers of time-shifting and region-shifting, engineers can optimize both costs and carbon emissions simultaneously. As regulations tighten and executives seek to reduce expenses while improving sustainability, GreenOps will likely become an increasingly important practice for cloud infrastructure management.