Disputing Incorrect PVU Values in an Audit Finding
PVU findings rest on two numbers IBM has to get right: the PVU per core rating for your processor, and the count of cores running the software. Both are frequently wrong, and both are contestable. Disputing them line by line is where findings shrink.
How a PVU finding is built
A Processor Value Unit charge is core based. IBM takes the PVU rating assigned to your processor model, for example roughly 70 PVU for a typical Intel core, and multiplies it by the cores allocated to the IBM software. If either input is overstated, the finding is overstated by the same proportion, and the error carries through to the price.
Where the per core rating goes wrong
The PVU per core value comes from IBM's processor table and depends on the exact processor model and generation. Auditors working from incomplete inventory data sometimes apply a higher rating than the hardware warrants, or assume a model that is not what is actually deployed. Confirming the real processor against the published rating is the first place a finding can move.
- A higher PVU per core rating applied than the actual processor model carries
- Processor generation misidentified from thin inventory data
- Full physical core counts used where sub-capacity should apply
- Cores counted on hosts where the software was not actually running
Where the core count goes wrong
The bigger swings usually live in the core count. If the deployment qualified for sub-capacity but the finding counts full physical capacity, the difference is dramatic. The setup primer puts it plainly: WebSphere using 4 of 32 host cores is 480 PVU under sub-capacity against 3,840 PVU at full capacity. Establishing the sub-capacity position, and proving how many cores the software truly used, is often the entire dispute.
Disputing it with evidence
Each PVU line is challenged the same way: confirm the processor model and its correct rating, establish whether sub-capacity applies, and reconcile the core count against your own ILMT or inventory evidence. Where the sub-capacity claim holds, the finding falls from full capacity to allocated cores. On average, challenges land 30 to 50% of findings, and PVU math is one of the most reliable places to recover.
A PVU finding is only as accurate as its two inputs, and IBM gets them wrong often enough to make every line worth checking. Confirm the rating, establish sub-capacity, and reconcile the cores against your own evidence. The wrong number is the opening position, not the settlement.