Skip to content
GreenKube

Cost Optimization

GreenKube combines FinOps and GreenOps to help teams find workload optimization opportunities. Recommendations are guidance based on collected metrics; GreenKube does not apply changes automatically.

Smart Detection

Section titled “Smart Detection”

GreenKube automatically analyzes your cluster to detect optimization opportunities:

🧟 Zombie Pods

Section titled “🧟 Zombie Pods”

Pods with cost above the configured threshold and energy consumption below the configured threshold.

📏 Rightsizing Opportunities

Section titled “📏 Rightsizing Opportunities”

Workloads where CPU or memory requests are substantially higher than observed use. The native analyzer suggests reductions; it does not diagnose workloads for increases as a rightsizing recommendation.

⚖️ Autoscaling Candidates

Section titled “⚖️ Autoscaling Candidates”

Workloads with variable CPU usage that have no existing HPA. GreenKube suggests considering autoscaling; it does not create or configure an HPA.

🌍 Carbon-Aware Scheduling

Section titled “🌍 Carbon-Aware Scheduling”

Recommendations identify workloads observed during higher-carbon periods as candidates for time-shifting. GreenKube does not move or reschedule workloads.

Savings Estimation

Section titled “Savings Estimation”

Where the recommendation type supports it, GreenKube annualizes projected cost and/or CO₂e savings from the observation window. These are projections, not measured results; cleanup recommendations may only include a cost estimate.

MetricDescription
Annual cost savingsProjected $ reduction per year
Annual CO₂e savingsProjected gCO₂e reduction per year
Priorityhigh, medium, or low

Recommendation Lifecycle & Savings Tracking

Section titled “Recommendation Lifecycle & Savings Tracking”

When you mark a recommendation as applied through the API, GreenKube records projected savings in its ledger. Post-apply verification can distinguish measured savings from prorated projections. The ledger drives Prometheus metrics including:

  • greenkube_co2e_savings_attributed_grams_total
  • greenkube_cost_savings_attributed_dollars_total

These appear in the Grafana Impact Command Center and on the dashboard summary.

How to Use

Section titled “How to Use”

Via Dashboard

Section titled “Via Dashboard”

Navigate to the /recommendations page to see active and ignored recommendations and savings. The dashboard supports GitOps pull-request previews and creation for eligible rightsizing recommendations. Marking a recommendation applied is available through the API.

Via API

Section titled “Via API”
Terminal window
curl "http://localhost:8000/api/v1/recommendations/active?refresh=true"
curl -X PATCH "http://localhost:8000/api/v1/recommendations/42/apply"

VPA and Karpenter are optional recommendation sources. Their Kubernetes CRDs are detected at runtime; absent CRDs are skipped. See the Recommendations Guide for details.

Integration with OpenCost

Section titled “Integration with OpenCost”

GreenKube reads cost allocation data from OpenCost, including pod-level data where available. Coverage and pricing depend on OpenCost’s configuration and the data it exposes.

Section titled “Related”