Sub-capacity licensing can cut a PVU bill by an order of magnitude, but it is conditional. The condition is evidence, and the evidence is the quarterly sub-capacity report produced by IBM License Metric Tool or another approved tool. If the report exists, runs continuously, and holds up, you pay for the cores the software actually uses. If it does not, IBM is contractually entitled to charge as though every physical core on the host runs the software. The report is therefore not a formality. It is the entire basis of the discount.
What the report actually proves.
PVU is a core-based metric. The full-capacity reading counts every physical core on a host and multiplies by the per-core PVU value, roughly 70 PVU for a typical Intel core. The sub-capacity report narrows that to the virtual cores allocated to the IBM product. The arithmetic is stark. A workload using 4 of a host's 32 cores is 480 PVU under sub-capacity and 3,840 PVU under full-capacity. The report is what entitles you to the smaller of those two numbers, and it has to do so for every measurement period, not just the moment of the audit.
The rules the report has to satisfy.
IBM's sub-capacity terms attach specific, non-negotiable conditions to that entitlement:
- Deploy within 90 days. The tracking tool must be installed within 90 days of the first eligible sub-capacity deployment.
- Run continuously. The tool must operate without interruption. Gaps in coverage void the sub-capacity claim for the periods they touch.
- Generate reports quarterly. A sub-capacity report has to be produced at least once per quarter.
- Retain reports for two years. Those quarterly reports must be kept for a rolling two-year window so they can be produced on audit.
Miss any one of these and IBM defaults the affected period to full-capacity charging. The conditions are cumulative, not a menu.
Why a report can be technically valid and still fail.
A report that exists is not automatically a report that defends you. ILMT routinely miscategorizes installs, which means the bundle a product is counted against can be wrong even when the tool is running. Broken or missing agents leave hosts uncovered, and an uncovered host is a full-capacity host for that period. A report generated late, or one that cannot be located for an earlier quarter, leaves a gap the auditor will price. The report has to be accurate and complete across the whole lookback, which can run two to five years.
Reading the report like an auditor.
When the audit arrives, IBM reads the sub-capacity report looking for the periods it can disqualify. A buyer side review reads it the same way first, to find and fix those gaps before they become findings. That means checking agent coverage against the full host inventory, confirming product categorization, verifying that a report exists for every quarter in scope, and reconciling the reported PVU against entitlements. The work is the same whether you do it or IBM does it. The difference is who controls the outcome.