ILMT configuration mistakes that cost millions.

The IBM License Metric Tool is the instrument that earns you sub-capacity licensing, and like any instrument it is only as trustworthy as its configuration. Most of the costly findings we see in IBM audits do not come from a tool that was never installed. They come from a tool that was installed, looked healthy on the console, and was quietly measuring the wrong thing the whole time.

The catalog miscategorization trap

ILMT relies on a software catalog to recognize what is installed. When the catalog reads a product incorrectly, the tool reports against the wrong entitlement, or against none at all. A bundled component counted as a standalone product, an edition the catalog maps to a higher tier, or a version string it does not recognize all produce numbers that look authoritative and are wrong. IBM treats the manual correction of these miscategorizations as your responsibility, so an uncorrected catalog error is your exposure, not theirs.

Partial coverage that reads as complete

Sub-capacity is only valid where virtual capacity is continuously measured. A scan group that excludes a subnet, an agent that was never deployed to a new cluster, or a host that drifted out of the scan schedule all create blind spots. The console stays green because it reports confidently on everything it can still see. The audit, however, tests the hosts the tool could not see, and for those hosts IBM is entitled to charge full capacity, counting every physical core whether or not the software ever touched it.

• Scan groups that silently exclude eligible hosts or subnets
• Agents missing on new or rebuilt servers that never re-enrolled
• Catalog versions left stale so new products are unrecognized
• Bundled components miscounted as separately licensable products
• Reports generated but never reviewed for accuracy or completeness

The quarterly report that was never kept

Sub-capacity also requires the quarterly PVU reports to be generated and retained for two years. A common and costly mistake is treating the dashboard as the deliverable and never producing or archiving the formal reports. When the audit asks for the report covering a quarter eighteen months ago and it was never saved, the period is unsupported regardless of how the tool looked at the time. The evidence has to exist as a kept record, not as a live screen.

Why the number gets large

Any single misconfiguration would be a bounded problem if it were caught the same quarter. The reason these mistakes reach into the millions is the audit lookback, which can reach back two to five years. A catalog error or a coverage gap that persisted unnoticed is re-priced at full capacity for every period it touched. The example IBM itself uses is instructive: a product using four of thirty two host cores is 480 PVU under a valid sub-capacity claim and 3,840 PVU at full capacity. Multiply that gap across years of unsupported periods and a quiet configuration error becomes a seven figure finding.