Optimize your infrastructure resources

Kubernetes cost optimization: compare requests, usage and capacity

Review Kubernetes pods, nodes and reserved resources to prioritize FinOps recommendations using the measurements actually available.

Try on my infrastructure

Data and limits to check

  • Kubernetes inventory: nodes, pods and workload context.
  • Requests, capacity and available measurements; runtime detail may require an agent.

Turn measurements into decisions

  1. Link inventory to available measurements over a representative period, including peaks and scheduled jobs.
  2. Compare allocated resources with observed usage. Review memory, business criticality and dependencies before reducing capacity.
  3. Review GO, PREPARE or KEEP recommendations, approve the change and check its effect. A recommendation does not replace your approval.

Decision example, without guaranteed savings

A workload reserves more CPU than it usually consumes. Check spikes, startup behavior, limits and placement constraints before changing requests. Savings also depend on capacity actually released across nodes.

Coverage depends on the connector, permissions and collected metrics. Inventory alone does not prove a machine can be downsized. Cost, energy and CO₂ remain estimates when the required inputs are not measured.