The Hidden Costs in Your Kubernetes Cluster You’re Probably Ignoring

If compute looks fine but the bill keeps climbing, it’s probably these hidden culprits.

• Zombie PVCs: orphaned volumes after deletes; audit with kubectl get pvc --all-namespaces.
• Idle load balancers: services scaled to zero but LBs still billing.
• Cross-AZ chatty services: egress between zones can exceed node cost.
• Noisy sidecars: logging/APM agents with unbounded buffers and high verbosity.
• Forgotten cron jobs: batch workloads running nightly with large requests.
• NAT and data transfer: NAT gateways and internet egress can dwarf compute for data-heavy apps.
• Snapshots you never delete: old backups and AMIs stick around; clean monthly.
• Over-provisioned requests: “temporary” bumps that never rolled back hide in waste.

Run a weekly hygiene job and delete what you don’t need. Your AWS bill will drop faster than most optimizations.***