Skip to content
GreenKube

Carbon Tracking

GreenKube converts modeled Kubernetes workload energy use into estimated CO₂ equivalent emissions using electricity-grid carbon intensity data (Scope 2) and hardware manufacturing data (Scope 3). These estimates are not a complete GHG Protocol inventory or a compliance report.

The Problem

Section titled “The Problem”

Engineering teams running Kubernetes workloads have no visibility into the carbon footprint they generate. Without measurement, reducing emissions is impossible — and cloud waste compounds the problem by creating unnecessary CO₂ on top of unnecessary spending.

GHG Scope Classification

Section titled “GHG Scope Classification”

GreenKube uses GHG Protocol Scope 2 and Scope 3 terminology for modeled operational and embodied emissions. Its workload estimates do not constitute a complete corporate inventory or compliance report.

Scope 2 — Operational Emissions (Electricity)

Section titled “Scope 2 — Operational Emissions (Electricity)”

Field: co2e_grams

Scope 2 covers the carbon emitted by generating the electricity your workloads consume.

co2e_grams = energy_kwh × grid_intensity_gCO₂e_per_kWh × pue

Data sources:

  • Energy: Estimated from CPU utilization × node power profile (see Energy Estimation)
  • Grid intensity: Real-time data from Electricity Maps API (default) or the free Wattnet API for European zones — selectable via config.electricityProvider
  • PUE (Power Usage Effectiveness): Provider-level data-center efficiency factors (AWS 1.15, GCP 1.09, Azure 1.18, OVHcloud 1.26, Scaleway 1.37)

Scope 3, Category 1 — Embodied Emissions (Hardware Manufacturing)

Section titled “Scope 3, Category 1 — Embodied Emissions (Hardware Manufacturing)”

Field: embodied_co2e_grams

Scope 3 Category 1 covers the upstream carbon cost of manufacturing the cloud hardware running your pods.

embodied_co2e_grams = (gwp_kg × 1000 / lifespan_hours) ×
                      (pod_duration_seconds / 3600) ×
                      (pod_cpu_cores / node_cpu_cores)

Data sources:

  • GWP (Global Warming Potential): Boavizta API per instance type (typically 50–170 kg for cloud VMs)
  • Fallback: DEFAULT_EMBODIED_EMISSIONS_KG = 100 kg when Boavizta has no data for a given instance type — calibrated to a cloud VM vCPU slice
  • Hardware lifespan: DEFAULT_HARDWARE_LIFESPAN_YEARS = 4 years (standard cloud equipment lifecycle)
  • Allocation: Proportional to the pod’s CPU share of the node

Total Footprint (Scope 2 + Scope 3)

Section titled “Total Footprint (Scope 2 + Scope 3)”

Field: total_co2e_all_scopes

This field combines modeled operational and embodied emissions. It is not a complete inventory of all cloud services or an audited organizational footprint.

How Carbon Zones Work

Section titled “How Carbon Zones Work”

Each cluster node is automatically mapped to an Electricity Maps zone based on its cloud provider and region. GreenKube includes built-in mappings for:

ProviderZone detection
AWSeu-west-1 → IE, us-east-1 → US-MIDA-PJM, etc.
GCPus-central1 → US-IL, europe-west3 → DE, etc.
Azurewesteurope → NL, eastus → US-MIDA-PJM, etc.
OVHGRA11 → GRA → FR, OpenStack nova zone handled gracefully
Scalewayfr-par → FR, nl-ams → NL, pl-waw → PL

If a zone can’t be resolved automatically, GreenKube falls back to a configurable default zone (DEFAULT_ZONE) or the global default intensity (DEFAULT_INTENSITY = 500 gCO₂e/kWh).

Using Carbon Estimates

Section titled “Using Carbon Estimates”

Use co2e_grams, embodied_co2e_grams, and total_co2e_all_scopes to compare workloads and track changes over time. Export historical data through the dashboard or API; see Reports & Exports. Keep the energy model, allocation assumptions, fallback values, and missing data alongside any exported totals.

Carbon Visibility Features

Section titled “Carbon Visibility Features”
FeatureDescription
Per-pod CO₂e breakdownSee exactly which workloads emit the most carbon
Scope 2 + Scope 3 per podModeled operational and hardware-embodied components in the estimate
Historical time seriesConfigurable multi-year retention
Estimation transparencyis_estimated flag + estimation_reasons per metric
Export formatsCSV, JSON, API access for BI integration

Configuration

Section titled “Configuration”

Set your Electricity Maps API token to use time-varying, zone-specific carbon intensity:

# Helm values.yaml
secrets:
  electricityMapsToken: "your-api-token"

Without a token, GreenKube uses static default intensities per zone from the bundled CSV. These estimates can help with trend analysis, but lack time-specific grid data. Get a free token at electricitymaps.com.

You can also select the free, EU-funded Wattnet provider for 15-minute-resolution grid intensity across 52 European zones:

# Helm values.yaml
config:
  electricityProvider: wattnet
secrets:
  wattnetEmail: "you@example.com"
  wattnetPassword: "your-password"

See the Wattnet guide for zone mapping and setup details.

Section titled “Related”