Db2 editions and their licensing metrics.

One product family, several editions

The Db2 family spans editions aimed at very different scales, from full enterprise database servers down to constrained or workgroup editions and developer or community offerings. They are not interchangeable for licensing. Each edition has its own entitlement basis and its own constraints on how much hardware it may use, and counting a deployment against the wrong edition is one of the most common ways a Db2 finding is inflated. The first reconciliation question is therefore always which edition is genuinely running, not which one the discovery tool guessed from installed files.

The metrics that apply

Db2 editions are licensed on a small set of metrics, and which one applies depends on the edition and how it was acquired:

• Processor value unit. The core-based metric where PVU per core, around 70 for a typical Intel core, is multiplied by the cores allocated to Db2. This is where sub-capacity matters most, because a virtualized deployment without clean ILMT evidence can be counted against the whole physical host rather than the cores it uses.
• Per core charge units. Newer entitlements may count cores directly under a virtual processor core style basis rather than PVU. Since Passport Advantage v11, VPC-metered products require ILMT based sub-capacity reporting and manual counting is no longer permitted.
• Authorized user. Some editions can be licensed per named user, often with a minimum number of users required per core, which means a small user count does not always mean a small license.

Where the edition and metric combination goes wrong

The errors compound because two variables are in play at once. An audit can assign the wrong edition and then apply the wrong metric to it, or apply a per user minimum incorrectly, or default a virtualized deployment to full capacity because sub-capacity evidence is missing. Each of these is a systematic error rather than a single line, so correcting it reverses the count across every affected install. The defense is to fix the edition first, confirm the correct metric for that edition and how it was entitled, and then, where the metric is core-based, prove the sub-capacity position from clean ILMT data. On a manufacturing estate running WebSphere and Db2, rebuilding the position this way against clean evidence and correct rates cut the claim by 71 percent.