A complete guide to interpreting your Kubernetes waste audit.
When pods have limits that are more than 2x their requests.
Example:
Cost: You're reserving 1.5GB of unused memory = ~$11/month wasted
When pods have CPU limits that are more than 3x their requests.
Example:
Cost: You're reserving 0.85 cores unused = ~$18/month wasted
PersistentVolumes that are provisioned but not attached to any pod.
Example: A 100GB PV sitting unused = $10/month wasted
LoadBalancer services with no backend pods (empty selector).
Cost: Each idle load balancer = ~$20/month wasted
Total money you could save per year by right-sizing resources. This is the headline number.
Same as annual, but divided by 12. Easier to compare against your monthly cloud bill.
Number of pods where limits are significantly higher than requests. These are your "quick wins."
Pods with no resource requests set. This is dangerous - they can starve other pods.
10-20%:Excellent! You're well-optimized.
20-40%:Typical. Most clusters have this much waste.
40-60%:High waste. Significant savings opportunity.
>60%:Critical. Review your resource settings ASAP.