6.11 Carbon Aware Development

Introduction:

The current carbon intensity of the electricity grid should drive the decision as to when energy-intensive processes, deployments, data transfers, file uploads and downloads, batch jobs are run. Do more when renewable energy is flooding the grid, do less when fossil fuels are dominant.

Not all datacentres are created equal either, even within the same “net-zero” provider. Placing resources in a particular region can meaningfully reduce environmental impact, for example, some Microsoft Azure regions carry Zero Waste certification, meaning at least 90% of food, office and construction waste is diverted from landfill through reuse, recycling and community partnerships. The Azure Global Infrastructure Map lets you compare regions before you choose where to deploy.

Tools such as ElectricityMap surface real-time grid intensity, and the open source Carbon Aware SDK is built specifically to support carbon-aware development decisions. If you’re running on Kubernetes, this can be operationalised directly: KEDA, the event-driven autoscaler, can now scale workloads based on carbon intensity via the carbon-aware KEDA operator, and Kepler: a project donated to the CNCF by Red Hat and announced at KubeCon Amsterdam, gives you carbon-based observability into your clusters.

N.B. this could be a server-side batch job (for example) or front-end adaptations.