Contents
Key takeawaysHow mainframe licensing worksThe rolling four hour averageSub capacity and SCRTTailored Fit PricingBroadcom CA renewalsWhat we saw in 2024 and 2025Renewal timelineWhat to do nextFAQIBM prices most mainframe software on MSU capacity, and Monthly License Charges follow the month's highest rolling four hour average. A short spike can set the whole month's bill, so controlling the peak usually saves more than any discount.
- MSU measures capacity. You pay for the capacity your workload peaks into, and MLC products bill against the highest rolling four hour average in the month.
- Control the peak first. In our work an uncontrolled peak inflated monthly charges 15 to 30 percent, and capping plus retimed batch removed that before any negotiation.
- SCRT reporting is the biggest avoidable gap. Without it the default is full capacity, and a missed monthly report costs a month of full capacity charges for that machine.
- Tailored Fit Pricing is a baseline year decision. Signed without modeling a low and a high year, it locked in 10 to 20 percent of avoidable cost.
- Broadcom CA renewals open high. Uplift asks of 20 to 40 percent softened sharply once third party support and tool retirement were on the table.
- Negotiate rate last. Peak control comes first, TFP modeling second, and the discount conversation only once the volume it applies to is right.
How is IBM mainframe software licensed?
Most IBM mainframe software is priced on Million Service Units (MSU), a measure of processor capacity. MSU works as a capacity meter rather than a usage counter. You pay for the capacity your workload peaks into, whatever volume it processes over the month.
Monthly License Charge (MLC) products such as z/OS, CICS, Db2, IMS and MQ bill every month against that peak. Tools under IBM's International Program License Agreement (IPLA) are bought once as a One Time Charge and then carry annual Subscription and Support. Four pricing models sit on top, each with its own place to control cost.
| Model | What you pay against | Where you control cost |
|---|---|---|
| MLC full capacity | The MSU rating of the whole machine | Reduce the machine, or move to sub capacity reporting |
| MLC sub capacity | The peak reported through SCRT | Cap the peak with defined capacity |
| Tailored Fit Pricing | An annual consumption baseline plus a growth allowance | Choose the baseline year deliberately |
| One Time Charge (IPLA) | A perpetual license plus annual support | Negotiate at purchase, when your position is strongest |
If you also hold an IBM Enterprise License Agreement, read it to see which mainframe products it covers and whether it fixes any MLC charges or leaves them on the MSU rules below. Our IBM ELA analysis covers the agreement structure above them.
The IBM Audit Is the Sales Call: Timing and ILMT Hygiene Decide It
How does the rolling four hour average set the monthly bill?
Under sub capacity MLC, IBM bills each product on the highest rolling four hour average (R4HA) of MSU consumption in the month, measured across the LPARs where that product runs. A short spike can set the bill for the whole month, so the timing of work matters more than its total volume.
The peak prices the month regardless of what the other 719 hours did. Batch work stacked onto the online peak is the classic self inflicted charge. Moving two batch windows off the online peak has been enough to drop the average, and the bill, before any negotiation opened.
Say an LPAR's online workload peaks at a rolling four hour average of 900 MSU on a normal weekday. On the last business day, month end batch starts at 16:00 while online traffic is still heavy, and the R4HA climbs to 1,150 MSU for a single afternoon.
If the MLC products run only in that LPAR, every one of them is billed at 1,150 MSU for the month. Start the batch at 20:00, after the online peak has passed, and the monthly peak falls back to 900 MSU. That removes 250 MSU, about 22 percent of the billed peak, with no change to the contract.
The dollar saving will be smaller than 22 percent, because MLC prices per MSU fall as you climb the pricing tiers and the MSUs removed at the top are the cheapest. It still recurs every month the new schedule holds.
Three things control the peak: the capacity available to each LPAR, defined capacity caps and workload placement. The two subsections below cover caps and placement.
How does defined capacity cap the chargeable peak?
Defined capacity is an LPAR setting, expressed in MSUs, held in the image profile on the Hardware Management Console. When the LPAR's rolling four hour average exceeds it, Workload Manager signals PR/SM to soft cap the partition, and the chargeable peak stays at or near the cap.
A Group Capacity Limit applies the same control to a group of LPARs on one machine. Caps only save money if the work behind them can wait, so set them with your performance team and test them against a month end run before relying on them.
Where do zIIP engines and workload placement fit?
Work dispatched to zIIP specialty engines does not count toward the MSU values that MLC products are billed on. Moving eligible Java, Db2 distributed and network work onto zIIPs lowers the general purpose peak. You can size the effect first with our MSU calculator.
The detailed reduction methods on the IBM side, workload shift, zIIP offload and the ILMT discipline for distributed software, are in our MSU reduction guide.
IBM MLC and IPLA negotiation brief
The pricing models, SCRT reporting, the TFP decision and the renewal sequence in one document.
Get the white paper →What is sub capacity pricing, and what happens without SCRT?
Under sub capacity pricing you license MLC products below the full capacity of the machine by reporting the actual measured peak through IBM's Sub Capacity Reporting Tool (SCRT). Without SCRT reports, the default is full capacity. On a large machine that is the largest avoidable cost in the mainframe budget.
SCRT reads SMF type 70 subtype 1 records for CPU activity and SMF type 89 subtypes 1 and 2 for product use, then produces a monthly report for each machine. IBM's timing rules are strict:
- Reporting period. The 2nd of each month through the 1st of the following month.
- Deadline. The report must reach IBM by close of business on the 9th day of the month.
- Missed report. IBM charges one month of full capacity for that machine.
Run the submission and the caps as one routine owned by one team, and reconcile reported peaks against your entitlements before every renewal. That way the baseline you negotiate from comes from your own data.
How do you check your own position?
- The last 12 SCRT reports. They show the MSU value reported for each product on each machine and when the peak fell. Line those peak times up against your batch schedule.
- RMF Partition Data reports. Built from the same SMF 70 data, they show each LPAR's four hour average MSU across the day.
- The HMC image profiles. These hold the defined capacity and group capacity values actually in force, which can differ from the capacity plan.
- The WLM service definition. It tells you which workloads get priority when a cap takes effect.
- Your MLC invoices. Match them month by month against the SCRT submissions, and flag any month billed at full capacity.
Which mistakes cost the most?
- A late or missing SCRT report. One missed deadline turns a sub capacity month into a full capacity month for that machine. Give the job a named owner and a backup.
- Caps set once and never revisited. A cap set too low shows up as delayed batch and slow online response, and one set too high saves nothing. Review each cap monthly against the SCRT report and write down who may raise it.
- Unchecked schedule changes. An application team that shifts a batch job into the afternoon can raise the monthly peak without anyone in licensing knowing.
- Negotiating rate on an unmanaged peak. Any discount then applies to capacity you did not need to buy.
Should you move to IBM Tailored Fit Pricing?
Move only after you have modeled both scenarios. Tailored Fit Pricing (TFP) replaces the monthly peak model with charges on the MSUs you actually consume over a year, set against a committed baseline plus a growth allowance. Because it measures consumption, IBM positions it as removing the need for capping.
That helps a steadily growing mainframe whose peaks you cannot control. It hurts a flat or shrinking one that would do better capping under the peak model. Treat TFP as a decision about which baseline year you commit to. Four points decide it:
- The baseline year. Model a low year and a high year before signing, because a baseline set in an unusually high year prices the whole term against it.
- The growth allowance. Compare the included growth to your real trajectory. Growth that runs below the allowance is consumption paid for and never used.
- The exit terms. Understand how you leave if consumption falls, before the model makes falling consumption expensive.
- The fit. The only way to know which side of the line you are on is to run both scenarios on the same history.
In the agreements we reviewed, TFP signed without a baseline year analysis locked in 10 to 20 percent of avoidable cost. The checks above are the ones those buyers skipped.
How does the TFP decision change with your situation?
| Your situation | Peak model with caps | Tailored Fit Pricing |
|---|---|---|
| Steady growth, peaks hard to control | Caps delay work the business needs done | Often better, with growth priced above the baseline and no capping |
| Flat workload | Caps keep the peak low at little cost | Risk of paying for a growth allowance you never use |
| Shrinking workload | The bill falls as the peak falls | The baseline commitment keeps charging for consumption you no longer have |
| Strong seasonal spikes | One spike prices the whole month | Annual entitlement spreads the spike across the year |
| New applications planned | New work raises the peak | Ask whether IBM's New Application Solution can price the new work separately |
What contract terms should you ask for in a TFP agreement?
- A named baseline period. Write the 12 months used and the data source into the contract, so the baseline cannot drift toward IBM's preferred year.
- A held growth price. Fix the price for MSUs above the baseline for the full term, since that is the rate you will pay as you grow.
- A right to reset the baseline. If a workload leaves the platform through migration or divestment, the committed baseline should fall with it.
- A written exit route. State how you return to sub capacity MLC, the notice required and any charges, before you need it.
- Clear reconciliation rules. Agree when the annual reconciliation runs, how consumption above entitlement is priced, and what happens to entitlement you did not use.
How are Broadcom CA mainframe renewals priced now?
The former CA Technologies mainframe tools, covering systems management, security, storage and development, now renew on Broadcom terms. The asks have been steeper since the acquisition, and 20 to 40 percent uplift arrived as the opening position across our renewals.
Those asks softened sharply once third party support options and selective retirement of low value tools entered the conversation. Broadcom's commercial approach across its portfolio tends to reward customers who arrive with alternatives.
What will the IBM and Broadcom account teams say, and how should you answer?
- IBM: "Tailored Fit Pricing gives you predictable spend and ends capping." Ask for the offer priced on your lowest and highest recent years, with the exit terms attached, and compare it with capped MLC on the same history.
- IBM: "Your SCRT data shows growth, so commit to a higher baseline while the growth price is good." Commit to the baseline your data supports and buy growth at a held rate when it actually arrives.
- Broadcom: "The uplift reflects our investment in the portfolio." Put your tool list on the table ranked by usage, name the tools you will retire, and show that you hold third party support quotes for the rest.
- Broadcom: "Consolidate everything into one portfolio agreement for a better price." Ask for each product priced separately first, with swap rights written in, so you can see what the bundle discount is worth.
The portfolio mechanics, the MSPS bundle, the swap rights and eight renewal tactics are worked through in our Broadcom CA mainframe pricing analysis. For the wider contract picture, see our Broadcom CA mainframe licensing guide for CIOs.
Broadcom prices on MSUs too, but usually on a different measure: traditionally the full MSU rating of the machine, or under its Mainframe Consumption Licensing program, the total z/OS MSUs consumed each hour as reported through SCRT. Capacity work can lower both vendors' bills, so check which measure each Broadcom order form uses before you count the saving twice.
What have we seen in IBM mainframe renewals in 2024 and 2025?
Across roughly 15 to 25 mainframe environments I advised for our clients between 2024 and 2025, the bill almost always moved more from capacity discipline than from discount negotiation. The same findings came up again and again:
- Peak control. An uncontrolled rolling four hour peak had inflated monthly charges 15 to 30 percent. Capping defined capacity and retiming batch cut MLC by the same band before the discount conversation opened.
- The TFP baseline. Consumption agreements signed without a baseline year analysis carried avoidable cost from the first month.
- Full capacity by default. Where SCRT reporting was missing or unreliable, the full capacity default was the largest avoidable cost on the bigger machines.
Why asking IBM for a bigger discount first gets the order wrong
The usual advice is to push for a bigger percentage discount and treat the MSU bill as fixed. We disagree, because in our work the discount moved the bill far less than capacity discipline did. A discount lowers the unit price of MSUs you did not need, and a controlled peak means you stop buying them.
The sequence that worked was peak control first, Tailored Fit Pricing modeled against a real baseline second, and rate negotiation last. Wherever we found rate treated as the opening request, the largest savings were still sitting untouched.
Every MSU kept out of the monthly peak comes off the bill for each MLC product in that partition, every month the new schedule holds.
When should you start preparing for an IBM mainframe renewal?
Start 12 months out. Peak control needs several monthly SCRT cycles before it shows in the reported numbers, and IBM prices any TFP offer from your history, so the year before renewal is the one that counts.
| When | What to do | Why |
|---|---|---|
| 12 months before | Collect 12 months of SCRT reports and MLC invoices, and confirm every machine reports by the 9th | You need your own baseline before IBM proposes one |
| 6 months before | Retime batch, set defined capacity and group caps, model TFP on a low and a high year, and list Broadcom tools by usage | Lower peaks have to appear in SCRT data before the renewal is priced |
| 3 months before | Reconcile reported peaks to entitlements, get IBM and Broadcom proposals line by line, and price third party support | You negotiate from your numbers and with alternatives in hand |
| 1 month before | Negotiate rate and settle the baseline, growth price, exit and swap terms | Rate comes last, once the volume it applies to is right |
What to do next
- Confirm sub capacity reporting on every machine. Check that each of the last 12 SCRT reports reached IBM by the 9th, and chase any invoice month billed at full capacity.
- Cap defined capacity and retime batch. Move batch windows off the online peak and set caps deliberately, before any negotiation starts.
- Model TFP against a low year and a high year. Do this before signing, and read the exit terms in the same pass.
- Reconcile reported peaks to entitlements. Finish this before renewal so the baseline you negotiate from is your own measured one.
- Price alternatives before the Broadcom CA renewal. Rank tools by usage, get third party support quotes and name the tools you will retire.
- Bring in help where you need it. Our IBM practice can run the SCRT review, the TFP model and the renewal with you.
Frequently asked questions
How is IBM mainframe software licensed?
Mostly on Million Service Units of processor capacity. MLC software such as z/OS, CICS and Db2 is billed each month on the peak rolling four hour average of the LPARs it runs in, while IPLA tools are bought once and carry annual Subscription and Support. The height of the peak drives the MLC bill.
What is the rolling four hour average?
It is the average MSU consumption over the previous four hours, recalculated continuously. IBM takes the highest value in the monthly reporting period for sub capacity MLC charges. Because it is an average, a brief spike raises it less than several hours of sustained load, which is why stacked batch and online work does the damage.
What is sub capacity pricing on the mainframe?
It is licensing MLC products on the measured peak of the LPARs where each product runs, instead of the full machine rating. You qualify by running SCRT and submitting its report every month. The saving is largest where a product runs in only some of the LPARs on a large machine.
Should a CIO move to IBM Tailored Fit Pricing?
Only if the modeling supports it. TFP suits steady growth with peaks you cannot cap without hurting service, while flat or shrinking workloads usually do better with caps under the peak model. Ask IBM to price both options on the same history, including what it costs to leave TFP later.
How do you handle Broadcom CA mainframe renewals?
Build alternatives before the quote arrives. List every tool with its real usage, get third party support quotes for the ones you keep, and name the ones you will retire. Opening asks come down when Broadcom sees you can walk away from part of the portfolio.
What cuts the mainframe bill most?
Capacity discipline. Capping the peak and retiming batch lowered MLC more than the discounts we negotiated, and that saving repeats every month the schedule holds. Rate talks belong after peak control and after TFP has been modeled against a real baseline.
What happens if an SCRT report is late?
IBM charges that machine at full capacity for the month. Each report covers the 2nd of one month through the 1st of the next and is due by close of business on the 9th. Keep the submission confirmations, because they are your evidence if an invoice later shows a full capacity month.