Db2 audit: the top five findings we see.

One: the wrong edition counted

Db2 ships in several editions with different entitlement bases, and discovery tools infer the edition from installed files rather than from what is licensed. The result is a deployment counted as a more expensive edition than the one genuinely running. This is the most common Db2 finding and often the largest, because the wrong edition usually carries the wrong metric with it. The fix is to confirm the real edition first, before any count is accepted, then apply the metric that actually belongs to it.

Two: sub-capacity defaulted to full capacity

Where Db2 runs virtualized and the ILMT evidence is incomplete, IBM defaults the count to the full physical host rather than the cores Db2 uses. A deployment using a handful of cores on a large host can be counted against every core in the box. Sub-capacity requires ILMT deployed within 90 days of first eligible deployment, running continuously, with quarterly reports retained two years. Where those conditions held, the full capacity default is reversible by producing the evidence.

Three: the bundled Db2 used beyond its scope

Db2 is bundled inside other IBM products, and those bundles carry restricted use rights. Using a Db2 that arrived bundled with another product for a workload outside the bundle's allowed scope creates a finding that looks like unlicensed Db2. The defense is to map each Db2 instance to how it was entitled, separating restricted use bundles from full Db2 entitlements, so that legitimate bundled use is not counted as a standalone license.

Four: authorized user minimums misapplied

Some Db2 editions can be licensed per authorized user, usually with a minimum number of users required per core. An audit can apply that minimum incorrectly, either ignoring it where it should raise the count or applying it where a different metric governs. Because the minimum interacts with core count, a small error in how it is applied compounds across the estate. Confirming the metric and the correct minimum for the specific edition resolves it.

Five: development and non production instances counted as production

Db2 carries development and test rights that differ from production entitlements, and discovery does not distinguish a developer instance from a production server. Non production deployments counted as production inflate the finding with capacity that was never licensable as production in the first place. Tagging each instance by its genuine role, and matching it to the right of use that applies, removes that inflation. On a manufacturing estate running WebSphere and Db2, rebuilding the position across exactly these kinds of errors cut the claim by 71 percent.