The Kubernetes Compass to Being More Sustainable
summary
How to make Kubernetes deployments more sustainable by aligning scheduling with renewable energy supply and reducing carbon and cost via temporal shifting (time-based scaling) and spatial shifting (multi-cluster placement). Concrete technologies described include Cube Green (autoscale-to-zero with patching), Crossplane control planes, KEDA combined with the Carbon Aware SDK and Carbon Intensity Exporter for time-driven scaling, and Carmada for multi-cluster placement; example results show a sample workload cutting emissions from ~53 kgCO2/day to 39 and then 27 kgCO2/day.
AI-written summary · about
chapters (15)
- 0:00Kubernetes sustainability
- 3:20Talk goals
- 4:21Data center energy impact
- 7:55Reduce idle workload waste
- 13:57Carbon intensity and renewables
- 16:27Carbon-aware demand shifting
- 19:19Kubernetes demand shifting architecture
- 24:56Carbon intensity data tools
- 27:09Temporal vs spatial shifting
- 27:32Temporal shifting with Keda
- 30:45Spatial shifting with Carmada
- 33:34Practical emission reductions
- 34:14Scheduling strategies
- 35:18Environmental and cost trade-offs
- 36:19Further reading
related talks
59mHow Kubernetes optimisation can combat climate change
47mHow Kubernetes optimisation can combat climate change
30mSustainability Chronicles: Innovate Through Green Technology
47mBuilding sustainable architectures with Kubernetes
14mOpen Source Ecosystem: Building Sustainable Solutions with Kubernetes
05mTowards a Sustainable World Realized by Kubernetes and Composable...
36mOptimizing Resource Usage in Kubernetes
12mKeynote: Cloud Native Sustainability - How You Can Help Right Now!