How to challenge an ALA capacity measurement

A capacity finding is a measurement, and a measurement is only as sound as the tool, the window, and the definition behind it. How to challenge an ALA capacity measurement is the practical question of how a buyer tests each of those three elements and reduces a capacity based shortfall to the figure the contract actually supports. The number on the page looks objective, but it rests on choices, and choices can be examined.

This field note sets out a method for challenging a Micro Focus capacity measurement, from the definition outward to the data. It pairs with our ALA and entitlement review track, and it builds on the broader treatment in capacity definitions in Micro Focus ALAs.

Start with the contractual definition

Before examining any data, establish what the authorization defines capacity to be: the resource counted, the unit, and whether the measure is sustained or peak. The definition is the standard against which the measurement must be judged, and a measurement that does not match the definition is not evidence of a shortfall, however precise it appears. A buyer who fixes the contractual definition first knows exactly what the measurement is supposed to show, which makes any deviation from that standard immediately visible.

Test the measurement tool

Ask what tool produced the figure and how it works. Capacity tools differ in what they sample, how often, and how they aggregate. A tool that records instantaneous peaks reports something very different from one that averages over an interval, and if the authorization defines a sustained measure, a peak reporting tool is measuring the wrong thing. The buyer is entitled to understand the tool's methodology, because a measurement produced by a method that does not match the definition cannot support the finding built on it.

Test the measurement window

Ask over what period the measurement was taken and whether that period reflects normal operation. A short window, or one that coincided with an atypical load event, exaggerates capacity. A representative window, covering ordinary operation over a meaningful span, reflects the sustained level the authorization more often licenses. Where the measurement window was short or unrepresentative, the figure overstates the licensable capacity, and a measurement taken over a proper window will commonly come in materially lower.

Separate burst from sustained

The most powerful single challenge is to separate transient bursts from sustained operation. Many systems spike briefly under load and settle far below the peak. Where the authorization defines a sustained measure, a finding built on the burst overreaches, because the burst is not the licensed level. Producing data that shows the sustained operation, distinct from the spikes, is frequently what moves a capacity finding the most. This is the same separation that reduced our E-03 case file, where an events per second and connector finding moved from $6.0M to $1.8M, a 70 percent reduction, after burst was split from sustained.

Account for virtualization and standby

On virtualized or clustered systems, ask whether the measurement counted the right surface. A figure that sums capacity across every node, including passive standby members, or that counts a physical host where the authorization counts an assigned partition, overstates the licensable total. Reading the capacity figure against the virtualization and standby provisions, examined further in ALA virtualization and partitioning rules, removes capacity the grant never placed in scope.

Distinguish provisioned headroom from measured use

Ask whether the finding counts what the hardware could do rather than what was measured. Systems are routinely provisioned with capacity above normal load to absorb growth and failover, and where the authorization licenses a measured capacity rather than the machine's physical maximum, that unused headroom is not a shortfall. Distinguishing provisioned capacity from measured, contractually relevant capacity removes a category of finding that counts potential rather than actual consumption.

Insist on a reproducible measurement

A capacity measurement that cannot be reproduced rests on data the buyer never agreed to accept. Insist on enough detail to reproduce the figure: the tool, the configuration, the interval, and the raw data. Where the measurement cannot be reproduced or does not match the contractual definition, the finding stands on an unexamined number. Requiring a transparent, reproducible measurement is legitimate and often decisive, because the burden of a credible measurement properly sits with the party asserting the shortfall.

Bring your own measurement

The strongest challenge pairs scrutiny of the vendor's figure with a measurement of your own, taken over a representative window against the contractual definition. A buyer that holds an independent, definition aligned capacity record enters the conversation with a credible alternative rather than only an objection. This is the Reconstruct discipline applied to capacity, and it shifts the discussion from the vendor's number alone to a comparison of two measurements, resolved on which one matches the definition.

Reduce to the defensible figure, then close it forward

Once the tool, window, and definition have been tested and the burst separated from the sustained, the finding reduces to the capacity the contract actually supports. The final step is to close the question forward: a forward agreement that fixes the capacity metric as sustained where appropriate, names the measurement method, and writes in the standby and virtualization treatment, so the next review begins from agreed definitions. Challenging the present measurement and closing the definition forward are two halves of the same work, and completing both keeps a capacity dispute from recurring.

For the full method, read the complete OpenText audit defense playbook, and for capacity defense across the Micro Focus estate see our ALA and entitlement review track. If a capacity measurement is driving your finding, open a case.