toolbox
Home / Startups & Digital Business / Reserved Instance Utilization Rate Estimator

Reserved Instance Utilization Rate Estimator

The Reserved Instance Utilization Rate Estimator measures how much of purchased reserved compute capacity is actually being used. It compares consumed reserved instance-hours with the total reserved instance-hours available during the same period, producing a utilization percentage plus the unused capacity that still carries a commitment cost. This is a practical FinOps metric for checking whether a reservation portfolio is sized close to steady-state demand.

A high percentage indicates that most committed capacity is being consumed, while a lower percentage points to reservations that may be idle, oversized, or mismatched to eligible workloads. The result does not by itself prove that a reservation is economical: utilization should be reviewed alongside the reserved discount, on-demand alternative, coverage, flexibility, and any exchange or modification options. Use the estimator with billing-export or utilization data that matches the same reservation scope and measurement period.

Inputs

units
hr
hr
Result
—
Reserved utilization rate
Available reserved hours—
Consumed reserved hours—
Unused reserved hours—
Unused capacity—

1. Enter reserved instance count
Use the number of reservations or equivalent reserved units included in the measurement scope.

2. Set available hours per instance
Enter the hours each reserved unit could have been used during the period.

3. Enter consumed reserved hours
Use eligible workload hours actually matched to the reserved capacity during that same period.

4. Keep the scope consistent
Do not mix one region, family, or billing period in the numerator with a different scope in the denominator.

5. Review utilization and unused capacity
Use both values to identify underused commitments and prioritize deeper reservation analysis.

Available reserved hours = Reserved instances × Available hours per instanceReserved utilization rate = Consumed reserved hours ÷ Available reserved hours × 100Unused reserved hours = max(Available reserved hours − Consumed reserved hours, 0)

Where:

  • Reserved instances — number of committed units in scope
  • Available hours per instance — potential usable hours for one unit during the period
  • Consumed reserved hours — eligible hours actually matched to the commitment

Assumptions: All reservations in the calculation are treated as equivalent units for the selected period. The estimator does not model size-flexibility normalization factors or provider-specific matching rules.

What the result means

The main result shows the share of available reserved capacity that was consumed during the measurement period.

If consumed hours exceed modeled available hours, verify that the numerator and denominator use the same normalized units and billing scope.

Given

  • 12 reserved instances
  • 720 available hours per instance
  • 7,344 consumed reserved hours

Calculation
Available reserved hours = 12 × 720 = 8,640 hours
Utilization = 7,344 ÷ 8,640 × 100 = 85.00%
Unused reserved hours = 8,640 − 7,344 = 1,296 hours

Result
85.00% reserved utilization

About 15% of modeled reserved capacity was unused during the period, so those reservations warrant a coverage and workload-placement review.

Is reserved utilization the same as reservation coverage?

No. Utilization asks how much purchased reserved capacity was used. Coverage asks what share of eligible workload usage was covered by reservations or another commitment.

Can utilization be above 100%?

With consistent equivalent units, it normally should not be. A value above 100% usually means the consumed-hours data and the available-capacity denominator were normalized differently.

Should stopped instances count as consumed hours?

Only if the provider bills or applies the reservation benefit to that usage under the commitment rules. Use billing data rather than instance state alone when possible.

Does a low utilization rate always mean I should cancel a reservation?

Not necessarily. Some commitments cannot be canceled, and flexibility, seasonal demand, or planned migrations may affect the decision. Treat the rate as a diagnostic metric, not an automatic action.

What period works best for this calculation?

Use a period long enough to smooth temporary spikes but short enough to catch structural drift. Monthly billing periods are common because cost and utilization data are usually reconciled that way.