Contents
Key takeawaysHow mainframe licensing worksHow one peak sets the billReducing costs without cutsSub capacity reportingTailored Fit PricingISV contractsWhat our reviews showedAnswering the account teamWhat to do nextFAQMainframe software is billed on measured MSU capacity, and the MLC invoice follows the busiest four hour interval of the month. Control that interval through scheduling, caps and accurate SCRT reports, and the charge falls before any discount is discussed.
- MSUs drive the cost. IBM Z software is priced on measured capacity in MSUs, with users and cores playing no part in the charge.
- MLC and IPLA need different work. MLC bills monthly on the rolling peak and responds to peak shaping, while IPLA behaves like perpetual licenses plus support and responds to capacity right sizing.
- Capacity engineering comes first. Batch retiming and soft caps cut the monthly bill 12 to 25 percent in our reviews, with no business function removed.
- Reporting errors cost money. A missing SCRT report means a month billed at full capacity, and smaller configuration errors leave savings unclaimed.
- Tailored Fit Pricing depends on its baseline. It helps spiky workloads, adds little for flat ones, and a baseline measured in an unusual year raises cost for the whole term.
- ISV contracts are the bigger half. Separately negotiated vendor contracts often cost more than the IBM platform software and are rarely benchmarked before renewal.
How is mainframe software licensed?
IBM Z software is priced on measured capacity in MSUs (millions of service units per hour), not on users or cores. Two charge models sit side by side on the same machine, and each responds to a different kind of cost work.
- Monthly License Charge (MLC). The operating system and the subsystems: z/OS, CICS, IMS, Db2 and MQ. It bills every month on the rolling capacity peak, so peak shaping lowers it.
- International Program License Agreement (IPLA). The tools and utilities. They carry an up front charge, optional sub capacity terms and an annual Subscription and Support fee, so they behave like perpetual licenses plus maintenance.
| Dimension | MLC | IPLA |
|---|---|---|
| Charge cadence | Monthly | Up front plus support |
| Driver | The rolling capacity peak | Licensed capacity |
| Typical products | Operating system and subsystems | Tools and utilities |
| How to lower it | Peak shaping | Capacity right sizing |
| Reporting dependency | SCRT for sub capacity | Optional sub capacity terms |
List which of your products fall on which side of that line before any cost work starts. Shaving a monthly peak cuts the MLC invoice from the next reporting period. An IPLA saving only appears when you decline to buy more capacity or trim the support base.
How does one batch peak set the whole monthly bill?
MLC charges follow the highest rolling four hour average (R4HA) of MSU consumption in the month, for the partitions where each product runs. One uncontrolled spike, usually overnight batch stacked on late online work, prices the entire month whatever the other intervals did.
The Sub Capacity Reporting Tool (SCRT) reads SMF type 70 and type 89 records to find that peak, over a reporting month that runs from the 2nd to the 1st. In 30 to 50 percent of the environments we reviewed, one batch window set the charge for the whole machine.
A worked example of peak shaping
Say a production partition group peaks at an R4HA of 1,000 MSU during the online day. At month end a batch stream starts at 22:00, overlaps with late online processing, and pushes the R4HA to 1,200 MSU for one night.
Assume, for illustration only, a marginal rate of $200 per MSU per month across z/OS and the subsystems. Your own rate is the top tier price on your last MLC invoice.
| Scenario | Peak R4HA | MSU removed | Monthly saving at $200 per MSU |
|---|---|---|---|
| Today, batch overlaps online work | 1,200 MSU | None | None |
| Discretionary batch moved to 02:00 | 1,080 MSU | 120 | $24,000 |
| Batch moved plus defined capacity of 1,050 | 1,050 MSU | 150 | $30,000 |
The last row is worth $360,000 a year for one schedule change and one capping setting. MLC prices are tiered, and the price per MSU falls in each higher tier. The MSUs you remove come off the top tier, the cheapest one, so a 12.5 percent MSU cut saves somewhat less than 12.5 percent of the invoice.
Capped batch also runs longer. Check its finish time against the morning online start before you set the cap.
IBM Z Mainframe Negotiation Brief
Charge models, capacity engineering, the Tailored Fit Pricing decision and a negotiation calendar for your next IBM Z renewal.
Get the white paper →How do you reduce mainframe software costs without removing function?
Profile the intervals that set the peak, move discretionary work away from them, cap what you can, fix the reporting, and only then negotiate. Each step depends on the one before it.
- Profile. Pull 12 months of SCRT reports and find the hours that set each monthly peak. You cannot shape a target you have not seen.
- Shift. Reschedule discretionary batch away from the online peak window, spreading load so no single interval spikes.
- Cap. Apply defined capacity limits that hold partitions below a chosen MSU level, set carefully so the bill is protected without starving work the business depends on.
- Fix the reporting. Confirm that sub capacity submissions are configured and on time every month, the most common reason achievable savings go unclaimed.
- Negotiate last. Open the discount conversation after the capacity profile is optimized, against a bill that is already smaller.
Soft caps and group capacity limits
Defined capacity is an MSU setting on the partition. When the R4HA exceeds it, Workload Manager has PR/SM soft cap the partition, and the charge cannot rise above that level.
A group capacity limit sets one ceiling across several partitions, so a busy partition can borrow headroom from a quiet one. Cap partitions where WLM service classes already rank online work above batch.
Moving work to zIIP engines
Work that runs on zIIP specialty engines does not add to the general purpose MSU figure that MLC bills on. Db2 distributed requests, parts of Db2 utilities and Java workloads are the usual candidates. The RMF workload activity report shows zIIP eligible time that still ran on general processors.
Why a better renewal discount is the wrong place to start
The usual advice is to wait for the IBM renewal and push for a larger discount on the MLC line. We disagree with that order, because a discount applies to whatever peak you hand IBM. In our reviews, moving one batch window often saved more than a year of discount negotiation.
What breaks sub capacity pricing, and how do you check yours?
Sub capacity pricing charges on the peak of the partitions where a product runs instead of the full rated capacity of the machine. It depends on clean, timely SCRT output. Reports are due by close of business on the 9th, and a month with no report is billed at full capacity for that machine.
Late submission is the obvious failure. Configuration errors do as much damage: in our reviews they left 10 to 20 percent of achievable savings unclaimed.
- Missing SMF 89 data. Products that write no type 89 records must be entered by hand in the NO89 section, or SCRT assigns them wrongly.
- Stale product lists. Retired products still reported on a partition keep drawing a charge.
- Default startups. A subsystem started by default on a test partition writes SMF 89 records and is billed there, even when no one uses it.
- Gaps in SMF collection. Missing intervals leave the report incomplete, can misplace the peak and invite questions from IBM.
Aggregation across machines
Grouping machines in one location can lower the effective per MSU rate, because the tiered prices apply to the combined MSUs. Machines in a qualified Parallel Sysplex can be aggregated under AWLC. Country Multiplex Pricing treats all eligible machines in one country as a single machine for sub capacity reporting.
Whether aggregation helps depends on how your workloads are spread, so model it on your own SCRT history.
How to check your own position
Start with the last 12 SCRT reports you sent IBM and the invoices that followed. Then work through this list.
- Peak hour per month. The SCRT report shows the hour that set each product's peak. Map it to your batch schedule.
- Defined capacity settings. Compare the image profiles on the Hardware Management Console with the peaks actually reached.
- Product to partition mapping. Confirm each billed product runs where SCRT says it runs.
- Submission dates. Find any month submitted after the 9th and check whether that invoice was billed at full capacity.
- zIIP eligible time on general processors. Read it from the RMF workload activity report.
Is IBM Tailored Fit Pricing worth it for your workload?
It can be for spiky, peak driven workloads, and it rarely is for flat ones. Tailored Fit Pricing replaces the rolling peak with a consumption baseline set from a historical measurement period and charges for growth above it. The single interval penalty disappears, and so does the need to cap.
IBM offers a Software Consumption Solution, measured on MSUs consumed and still reported through SCRT, and an Enterprise Capacity Solution priced on the size of the physical environment. Flat workloads gain little from either, because their peak and their average already sit close together.
The baseline is where the risk sits
A baseline measured during an atypical peak season locks higher cost for the whole term. Model the current charge model against TFP on a low year and a high year before committing.
Ask IBM which months form the baseline, how growth above it is priced, and what happens to the baseline if you move workload off the platform during the term.
The schedule moved the bill more than the negotiation did, because the interval that sets the charge is under your control.
Why do mainframe ISV contracts deserve as much attention as IBM?
They usually cost more than the IBM platform software. Independent software vendor contracts on the mainframe are negotiated separately from IBM, and in our reviews they ran 1.5 to 3 times the platform software cost. They were also renewed without benchmarking.
A cost program that stops at the IBM invoice has worked on the smaller half of the budget. Most ISV products price on the same measured MSUs, so peak shaping can lower them too, provided each contract measures the capacity you actually use.
Contract terms to ask for at the next ISV renewal
- A sub capacity metric. Fees based on the MSUs of the partitions where the product runs, so your shaping work reduces them.
- No upgrade fee on hardware refresh. A new machine with more installed capacity should not trigger a charge if measured usage has not grown.
- A cap on renewal increases. A fixed ceiling on annual maintenance increases for the full term.
- Reduction rights. The right to drop products or capacity at renewal as workloads leave the platform.
- A clear notice date. Written into the contract, so an automatic renewal cannot pass before your benchmark is done.
The largest ISV renewal for most mainframe shops is Broadcom's CA portfolio, covered in our Broadcom CA pricing analysis and the CA mainframe licensing guide for CIOs.
What have we seen in mainframe cost reviews since 2024?
Across roughly 20 to 30 mainframe environments we reviewed between 2024 and 2025, capacity engineering cut the monthly software bill 12 to 25 percent without removing any business function. The median MSU bill cut from peak shaping alone was around 20 percent, before any negotiation opened.
The ISV contracts ran about twice the platform line at the median. The platform account team's standard position was that mainframe cost is essentially fixed and only a periodic renewal discount brings it down. In these reviews that position failed in a measurable way, and the MSU figures on the invoices show it.
What will the IBM account team say, and how should you answer?
Three lines come up in most mainframe renewals. Each has a reply you can support with your own data.
"Your mainframe software cost is essentially fixed."
Reply with your SCRT history. Show the hours that set each monthly peak and the MSUs removed since, then ask IBM to price the renewal on the new profile.
"Tailored Fit Pricing will make your costs predictable."
Ask for the proposed baseline months in writing, the price for growth, and a side by side model on your own history. Predictable only helps if the baseline was set in a normal year.
"Capping puts your service levels at risk."
Caps sit on partitions where WLM already protects online work, with headroom above the normal daytime peak. Offer the RMF data showing service class goals were met during capped intervals.
When to start before a renewal
Peak shaping needs several months of measured results before IBM will accept the new profile. Work back from the renewal or TFP decision date.
| When | What to do |
|---|---|
| 12 months before | Collect 12 months of SCRT reports and invoices. List every ISV contract with its metric and notice date. |
| 9 months before | Profile the peak hours and start moving discretionary batch. |
| 6 months before | Set defined capacity or group limits and measure the new peak. |
| 3 months before | Run the TFP comparison. Benchmark the ISV quotes. |
| 1 month before | Negotiate on the reduced profile, with the contract terms in writing. |
The operational detail of workload shift, zIIP offload and ILMT reporting for Linux on IBM Z sits in our MSU reduction guide, and the mainframe MSU calculator estimates your own figures. The strategy layer above them is in the mainframe CIO advisory.
What to do next
- This month. Profile the intervals that set your rolling peak in each of the last 12 SCRT reports.
- Next quarter. Shift discretionary batch off those intervals and apply soft caps where WLM protects online work.
- Every month. Verify that the sub capacity report is complete, correct and submitted by the 9th.
- Before any TFP decision. Compare both charge models on your quietest and busiest recent years, and get the baseline months in writing.
- Before each ISV renewal. Benchmark the price and ask for a sub capacity metric, a renewal cap and reduction rights.
- When you want help. Our IBM practice runs this program with you, from SCRT review to the final negotiation.
Frequently asked questions
How is mainframe software licensed?
On measured capacity in MSUs under two IBM models. Monthly License Charge software, meaning z/OS and the subsystems, is billed each month on the rolling capacity peak. International Program License Agreement products, the tools and utilities, are bought up front with optional sub capacity terms and an annual support fee. Peak shaping lowers the first, capacity right sizing the second.
How do you reduce mainframe software costs?
Shape capacity instead of removing function. Find the hours that set the rolling peak, move discretionary batch out of them, and cap partitions where the workload allows. In our reviews the median MSU bill cut from that work was around 20 percent, and it came before the renewal discussion started.
What is the rolling four hour average?
It is the average MSU consumption of a partition over the preceding four hours, recalculated through the month from RMF data. For MLC products IBM bills on the highest value it reaches in the reporting month, which runs from the 2nd to the 1st. A single busy evening can therefore set the charge for every day of that month.
What is sub capacity pricing and what breaks it?
It means paying on the peak of the partitions where a product runs rather than on full machine capacity. It breaks when SCRT reports are late, incomplete or mapped to the wrong partitions. A missed report reverts that machine to full capacity for the month, and misconfiguration left 10 to 20 percent of achievable savings unclaimed in our reviews.
Is IBM Tailored Fit Pricing worth it?
For spiky, peak driven workloads it can be, because consumption pricing removes the single interval penalty. Flat, predictable workloads gain little. The risk is the baseline: if the measurement window fell in an unusually busy season, you pay for that season all term. Compare both models on a low year and a high year first.
How big are mainframe ISV costs compared to the platform?
They are usually larger. In the environments we reviewed, separately negotiated ISV contracts cost roughly twice the IBM platform software at the median and were renewed without benchmarking. Most of them price on measured MSUs, so the peak work you do for IBM can lower them too, provided each contract uses a sub capacity metric.
Can one batch job really set the whole month's mainframe bill?
Yes. Because MLC charges follow the highest rolling average of the month, one overnight batch run stacked on late online work can price the whole month. That happened in 30 to 50 percent of the environments we reviewed, and rescheduling a single batch window often saved more than a year of discount talks.