Non production capacity was licensed at full production rates in a majority of estates, often doubling avoidable cost with no compliance benefit at all
Development is not production and disaster recovery is not always active. The default treatment is the expensive one, and it has to be challenged deliberately.
Prepared by Redress Compliance · August 19, 2026 · IBM environment optimization engagements. 25 to 35 engagements run, 2024 to 2025.
Executive summary
Non production capacity was licensed at full production rates in a majority of estates, often doubling avoidable cost by inertia rather than by requirement.
Sub capacity reporting covered production but not non production, and sub capacity reverts to full capacity on any cluster the tooling does not reach.
Cold standby and disaster recovery nodes were entitled identically to active production, with no reduced treatment applied anywhere in the estate.
The median entitlement reclaimed was 35 percent. None of it required a concession, only a classification the estate had never made.
How is non production actually licensed?
It depends on the program. Where a separate non production right exists it is materially cheaper than production entitlement, and where it does not, the default applies.
The default is the expensive path
Do nothing and non production consumes full entitlement at production rates. That is the costly outcome, so it has to be challenged deliberately rather than discovered at the renewal.
- Non production terms: apply where the program offers them.
- Development rights: some products bundle limited development use.
- Default: full production entitlement unless changed.
Check the program, do not assume the rule
Program specific terms are published on Passport Advantage, and the program documentation is where the entitlement question is settled rather than argued.
How does sub capacity apply to non production?
Exactly as it applies to production: entitlement drops to the cores actually used, and it requires the measurement tooling everywhere.
| Environment | Default treatment | Optimized treatment | The lever |
|---|---|---|---|
| Production | Full entitlement | Sub capacity | Measurement tooling |
| Test and staging | Full production rate | Non production terms | Program rights |
| Development | Full production rate | Development entitlement | Program rights |
| Cold standby | Full entitlement | Reduced standby right | Backup terms |
Everywhere or nowhere
Sub capacity reverts to full capacity on any cluster the tooling does not cover, so deploying it across production alone holds none of the saving on the environments that generate the overspend. The reporting rules sit in the sub capacity licensing terms.
The IBM audit defense playbook
Non production and development entitlement rules, sub capacity posture, standby treatment and the buyer side moves across the estate.
Get the brief →What 25 to 35 IBM environment engagements showed
Across roughly 25 to 35 IBM environment optimization engagements Morten Andersen ran between 2024 and 2025, development and test environments were repeatedly licensed as if they were production. Three patterns recur.
- Non production capacity was licensed at full production rates in a majority of estates, often doubling avoidable cost.
- Measurement tooling was deployed on production but not consistently across non production, forcing full capacity there.
- Cold standby and disaster recovery nodes were entitled identically to active production with no reduced treatment applied.
Most IBM overspend on environments is invisible, because development, test and staging are treated as just more capacity.
- Your agreements decoded into plain English before the auditor interprets them for you
- Entitlements, caps and protections verified across your whole contract portfolio
- A defensible position paper generated in minutes rather than weeks
How should standby and disaster recovery be licensed?
Node by node, according to how active each one actually is. Entitling the whole tier as production is the shortcut that costs the most.
Cold, warm and active are three different answers
- Cold standby: reduced or no entitlement when the node is idle.
- Warm standby: treatment closer to production.
- Disaster recovery: apply the backup terms per product.
Container platform products bring the same environment question with them, worked through separately in the data platform licensing guide.
Classification is the whole exercise
The treatment is product specific, so it has to be applied per node rather than per tier. A node classified once and never revisited is the version of this that quietly renews at production rates for years.
Which buyer side moves work on IBM environments?
Three recur: classify the environments, extend the measurement, and reconcile the entitlement against both.
Classify every environment first
Tag each cluster as production, test, development or standby and apply the right terms to each. Nothing downstream works until that map exists.
Extend the measurement to non production
Cover non production clusters so sub capacity holds across the whole estate rather than on the half that was already cheapest to report. The tooling documentation is at the metric tool reference, and the audit posture sits in the audit defense reference.
Where the common advice on non production is wrong
The common advice is to license every environment the same way for simplicity, so compliance is never in doubt. We disagree.
Uniform production licensing is not a compliance strategy
In most of the IBM estates reviewed, licensing development, test and standby at full production rates created large recurring overspend with no compliance benefit at all.
The buyer side move is to classify every environment, apply non production and development terms where the program allows, extend the measurement across non production, and treat cold standby under backup terms. It is a standing overpayment for environments that never carried a production workload. The agreement mechanics sit in the enterprise agreement reference.
What the engagements measured, 2024 to 2025
Two cuts of the engagement file, both recoverable by classification alone.
Against a classified estate, where development and test carried production entitlement by inertia rather than by requirement.
After environments were classified, non production terms applied and the measurement extended across the whole estate.
Neither figure came from a discount. Both came from licensing each environment for what it is.
Your first five moves
- Inventory every environment by type across the estate, because the classification map is what every other move depends on.
- Confirm the measurement tooling covers non production clusters, since sub capacity reverts to full capacity on any cluster it does not reach.
- Apply non production and development terms wherever the program allows them, which is materially cheaper than production entitlement.
- Classify standby and disaster recovery nodes individually, because cold, warm and active carry three different treatments and the tier answer is the expensive one.
- Reconcile entitlement against environment and measured use. The IBM practice and the spend health check run the environment reconciliation before the renewal opens.
Frequently asked questions
Is non production licensed differently?
It depends on the program. Where a distinct non production or development right exists it is materially cheaper, and where it does not, full production entitlement applies.
What happens if nothing is done?
Non production consumes full entitlement at production rates. The default is the costly path, which is why it has to be challenged deliberately rather than discovered.
How large is the overspend?
Typically about twice what a classified estate pays. Non production carried production entitlement in a majority of the estates reviewed.
Does sub capacity apply to dev and test?
Yes, exactly as it applies to production. It reduces entitlement to the cores actually used, and it requires the measurement tooling on those clusters too.
What if the tooling is production only?
Sub capacity reverts to full capacity on any cluster the tooling does not cover, so a production only deployment holds none of the saving where the overspend actually sits.
How is cold standby treated?
A node that is off until a failover may carry reduced or no entitlement. A warm or active node is treated closer to production, so classify each node rather than the tier.
How is disaster recovery handled?
Under the backup terms for each product, applied node by node. The treatment is product specific, so entitling the whole tier as production overpays by default.
How much gets reclaimed?
A median of 35 percent of entitlement after classification. None of it required a vendor concession, only a map the estate had never drawn.
Is uniform licensing safer for compliance?
No. It created large recurring overspend with no compliance benefit at all. It is a standing overpayment for environments that never carried a production workload.
Where should the work start?
With an inventory of every environment by type. Nothing downstream, from terms to measurement to reconciliation, can be aimed without that map.
