Public cloud eligible environments for IBM sub-capacity.

Public cloud feels like it should solve sub-capacity by design. You rent a small instance, you pay for a small instance, so surely you license a small instance. IBM does not see it that way unless three conditions are met at once: the provider is approved, the instance configuration is eligible, and the consumption is reported through an approved tool on a continuous basis. Each condition is a separate point of failure, and a finding only needs one to fail to revert that environment to full-capacity charging.

The principle is the same one that governs sub-capacity everywhere. IBM grants relief in exchange for evidence. In a data center the evidence is ILMT watching a known host. In public cloud the host is abstracted, so the eligibility rules narrow what counts as a measurable, approved environment, and the reporting rules require that measurement to be unbroken.

The three conditions

• Approved provider. Sub-capacity in cloud is recognized only on the cloud providers IBM lists as eligible. A workload on an unlisted provider does not get cloud sub-capacity treatment.
• Eligible instance type. Within an approved provider, only certain instance families and configurations qualify. Capacity is counted against the virtual cores of that instance, so the instance choice sets the number.
• Continuous reporting. ILMT or an approved tool such as Flexera One ITAM or HCL BigFix Inventory must measure the environment and produce reports continuously, retained for the standard period. A reporting gap is a full-capacity period.

Where eligibility breaks

The common failures are quiet ones. A team spins up the right software on a convenient instance family that happens to be ineligible. An environment moves region or resizes and the reporting tool loses sight of it. A short-lived autoscaling fleet runs IBM software on nodes that were never inventoried. None of these look like compliance events at the time, but each one removes a piece of the evidence chain, and IBM rebuilds the missing pieces at full capacity.

How to prove eligibility holds

The defensible position is a per workload map: which product, on which provider, on which instance type, measured by which tool, with reports retained. Where the workload uses the VPC metric, the count follows the virtual cores of the eligible instance. Where it still uses PVU, the per core rating applies to those virtual cores. Either way, the claim is only as strong as the continuous record behind it, so the reporting must be treated as a standing operation rather than something assembled when the audit letter arrives.