Sustainability

Sustainability is the estimate of the CO2 that hibernation avoids, and the UI page that shows it. To make the estimate, Hibernator records which instance types your nodes run. It uses the static coefficients of Cloud Carbon Footprint (CCF). Every value named here is in the table in configuration.md, under the values key config.insights.

Every hour, the controller samples the resource states and the node instance types. It adds the samples up into daily and monthly summaries, which the API serves at GET /api/v1/insights.

Sustainability is on by default, with the coefficients for eu-central-1 (Frankfurt). For a cluster in another region, set gridIntensityGCO2ePerKWh to the value of that region (see Region examples).

Configuration

config:
  insights:
    enabled: true
    pue: 1.14                               # AWS data center PUE (2025 disclosure)
    gridIntensityGCO2ePerKWh: 298           # Frankfurt grid (DE 12-month avg)
    cpuWattsPerVCPU:
      amd64: 2.30                           # Cascade Lake/Sky Lake at 50%
      arm64: 1.084                          # Graviton2 at 50%
    memoryWattsPerGB: 0.392

Region examples

Region	Grid intensity (gCO2e/kWh)	Source
eu-central-1 (Frankfurt)	298	DE 12-month avg (Jun 2025-May 2026)
eu-west-1 (Ireland)	296	EEA / CCF
eu-north-1 (Stockholm)	8	EEA / CCF
us-east-1 (N. Virginia)	379	EPA eGRID / CCF
ap-southeast-1 (Singapore)	408	IEA / CCF

Limitations

• The controller samples once an hour, so it can count a hibernation cycle shorter than one hour too high or too low.
• The estimate uses the resource requests of the pods, not their real use.
• The grid intensity is a static annual average, not a real-time value.
• Fargate pods use averages per architecture, because they have no instance type to map.