Decide whether an apparent saving survives billing reconciliation, application demand, interruption, and a one-node failure. Separate a financial rate change from the ability to acquire capacity when it is needed.
Review these lessons
- Shared cluster cost: reconcile service allocation with the provider bill
- CPU request rightsizing: account for the HPA feedback loop
- Interruptible compute: price recovery work, not only cheap node hours
- Compute commitments: separate billing coverage from machine availability
- Cross-zone data paths: measure bytes before changing placement
- NAT and private endpoints: compare the complete route and failure domain
- Cost alerts: account for billing delay before a runaway resource spreads
- Node consolidation: calculate the capacity needed to evict safely
Other checks
Common Mistakes
- Do not call unallocated cost zero.
- Do not make an availability decision from a discounted price alone.