Definition
Percentage of total slice resource-hours that were allocated but not consumed by actual demand, measuring the cumulative cost of the lifecycle model's over-provisioning bias.
The Measurement Gap
Measurement Gap and Decision
Decision Supported: How much of the slice infrastructure budget is being consumed by slices that exist but are not needed, and is the over-provisioning bias getting better or worse over time?
Measurement Gap:
Network slicing platforms report resource utilization at the aggregate level. Total allocated capacity versus total consumed capacity across the slice portfolio. This aggregate view hides the per-slice waste pattern. A portfolio can show 75% aggregate utilization while individual slices sit at 20% utilization for hours before demand arrives. The aggregate number also nets over-provisioned slices against under-provisioned slices, making the waste invisible. Operators track slice count and total resource allocation but do not track cumulative idle resource-hours per slice: the actual unit of waste. SOWI measures this directly: for every slice in every measurement interval, how much allocated capacity went unused? The sum across all slices and intervals, expressed as a percentage of total allocation, reveals the true cost of the lifecycle model's provisioning bias. This connects a technical metric to a financial impact: SOWI multiplied by the unit cost of slice resources equals the direct dollar cost of over-provisioning.
Citations:
- TM Forum, "Autonomous Networks: Technical Architecture," TR290, 2024. Describes intent-based lifecycle management but provides no waste measurement framework for slice-level resource allocation.
- Heavy Reading, "Economics of 5G Network Slicing," 2023: quantifies the business case for slicing but does not define metrics for measuring provisioning waste at the slice level.
Formula
Plain Language: For each active slice in each measurement interval, compute the ratio of actual demand to allocated capacity. Sum the unused capacity across all slices and all intervals. Express as a percentage of total allocated capacity.
Notation:
SOWI = [Σ_i Σ_t (Allocated_capacity_i,t − Demand_actual_i,t)] / [Σ_i Σ_t Allocated_capacity_i,t] × 100
Where the sum only includes intervals where Allocated > Demand (do not net out intervals where demand exceeds allocation).
Variable Definitions:
- Allocated_capacity_i,t = resource capacity allocated to slice i during time interval t (resource units: Mbps, CPU cores, or normalized capacity units)
- Demand_actual_i,t = actual demand on slice i during interval t (same units as allocated capacity)
- Σ_i Σ_t = summed over all active slices i and all time intervals t in the measurement period
- Only positive differences (over-provisioned intervals) are summed: under-provisioned intervals are a separate problem measured by SLA compliance metrics
Units: Percentage. 0% = every allocated resource-hour was fully consumed (perfect efficiency). Higher values indicate more waste.
Sign Convention:
Higher SOWI is worse. The metric only captures over-provisioning waste; under-provisioning is excluded by construction.
Edge Cases:
- Slices with guaranteed minimum allocations that sit idle as standby (e.g., public safety slices): exclude or segment separately. These are intentionally over-provisioned for availability reasons.
- Shared/elastic resource pools where "allocated" is a soft reservation: use the guaranteed minimum as the allocation baseline, not the elastic maximum.
- Minimum sample: computed across all active slices. Report by slice type for actionable segmentation.
Proposed formula, pending validation pilots.