Tailored Fit Pricing fixes your price per MSU from a trailing twelve month baseline, then carries it for the multi year term. Optimize MSU first and a representative 2,600 MSU estate cuts its IBM Z software stack from $6.0M to $4.5M a year.
Prepared by Redress Compliance · June 2026 · Representative IBM Z mainframe estate scenario (benchmark scenario, not a quote).
The IBM Z mainframe is the most expensive single platform in most enterprise estates, and its software, not its hardware, carries the bigger bill. The software splits into two worlds, MLC monthly license charges for z/OS and the major subsystems, and IPLA one time charge programs with annual support.
Both worlds meter on the MSU, a unit of consumed processing capacity. Sub capacity MLC bills on the rolling four hour average peak MSU, not the instantaneous peak, which is the single most important and least understood mechanic on the platform.
Tailored Fit Pricing, introduced by IBM in 2019, sets a price per MSU from your prior twelve months of consumption, then runs it forward with a contracted growth band. The migration is the most consequential commercial event in the IBM Z relationship, and the trap is that it grandfathers whatever consumption you bring to the table.
In the representative estate, the vendor shaped path totals $6,000,000 a year across MLC and IPLA. Optimizing MSU before the baseline, then negotiating the term, lands the same workloads at $4,500,000, a $1,500,000 saving and a 25 percent cut.
This paper decodes the commercial model, the TFP migration, MSU optimization, the Sub Capacity alternative, the IPLA portfolio, the z16 to z17 hardware refresh, and the seven contract levers that hold IBM accountable through the term.
The IBM Z software bill is two separate licensing systems sharing one consumption metric, not a single price. MLC governs the operating system and the major subsystems on a recurring monthly charge. IPLA governs tools and middleware on a one time charge plus annual support.
The unit underneath both is the MSU, Million Service Units, IBM's measure of processing work. The machine reports MSU consumption, the Sub Capacity Reporting Tool aggregates it, and the peak figure drives the invoice.
| Element | Licensing system | Meters on | Buyer note |
|---|---|---|---|
| z/OS | MLC, sub capacity AWLC | Rolling four hour average MSU | The anchor charge. Soft capping its R4HA moves the whole bill. |
| CICS, Db2, IMS, MQ for z/OS | MLC, sub capacity AWLC | Rolling four hour average MSU per product | Each product peaks on its own LPAR set. Tune the peaks separately. |
| Db2 tools, OMEGAMON, utilities | IPLA one time charge plus S&S | Full or sub capacity Value Units | Already paid for. Annual S&S is the recurring lever, not relicensing. |
| Specialty engines (zIIP) | Hardware feature | Offloaded eligible work | Work on a zIIP does not count toward billable general purpose MSU. |
| Tailored Fit Pricing | MLC replacement model | Baseline MSU plus growth band | Fixes price per MSU from trailing twelve months. Set the baseline deliberately. |
The non obvious mechanic is that MLC and IPLA do not move together. You can be heavily over consuming on z/OS MLC while paying flat S&S on IPLA tools you no longer run. A blanket percentage discount across the agreement hides both, so each system needs its own reconciliation.
The second mechanic is the four hour average. Sub capacity MLC does not bill the instantaneous peak. It bills the highest rolling four hour average of MSU in the month, which means a short spike costs nothing if the surrounding hours are quiet, and a sustained midday plateau costs everything.
You migrate to TFP only after you have optimized the consumption that sets the baseline, never before. TFP reads your trailing twelve months of MSU and MLC charges, derives a price per MSU, and runs it forward with a negotiated growth allowance for the term.
The baseline year selection is the entire game. A baseline drawn from a quarter end heavy, unoptimized twelve months locks a high price per MSU into a multi year contract. The same estate, soft capped and tuned first, sets a materially lower baseline.
The standard IBM and reseller pitch is that TFP is simpler and cheaper, so migrate at the next renewal and stop worrying about MSU. We disagree.
In the IBM Z estates we benchmarked in 2024 to 2025, customers who migrated first and optimized later handed the technology dividend and every soft capping gain to IBM, because the baseline had already fixed the price.
The buyer move is to reverse the order. Run the MSU reduction program, let consumption settle for a full quarter, and only then set the TFP baseline. Optimize first, sign second.
You optimize MSU by managing the rolling four hour average peak down, because that peak, not total work, sets sub capacity billing. The lever is workload placement and capping, not running less business.
Across the IBM Z estates we reviewed in 2024 to 2025, billed R4HA ran 20 to 35 percent above sustained workload once batch windows and soft capping were tuned. That gap is the headroom a TFP baseline would otherwise grandfather.
Figure 1. A 2,600 MSU billed peak resolves to 1,750 sustained, a 350 burst buffer, and 500 reclaimable, so the baseline target falls to 2,100 MSU. Benchmark scenario, not a quote.
Legacy Sub Capacity Licensing remains the better choice when your workload is flat or declining, because you keep banking the technology dividend that a TFP baseline would freeze. TFP rewards growth, sub capacity AWLC rewards shrinkage.
The decision turns on the shape of your roadmap, not on which model IBM is promoting this year. Both are valid, and the right answer differs by customer profile.
| Model | Bills on | Best fit | Watch for |
|---|---|---|---|
| Sub capacity AWLC | Monthly R4HA peak per product | Flat or declining MSU, active capping | Discipline cost. Every month must be measured and managed. |
| TFP Software Consumption | Baseline MSU plus metered growth | Predictable, modest growth roadmap | Baseline grandfathers the consumption you bring. |
| TFP Enterprise Capacity | Committed capacity band | Aggressive growth, new workloads | Commit floor. Pay the band whether you use it or not. |
| Container Pricing | Ring fenced new workload | Net new applications, payments, dev and test | Scope creep. Keep the container boundary tight. |
The contrarian point most account teams skip is that a shrinking estate is penalized by TFP. If you are decommissioning applications or moving workload off platform, the legacy sub capacity model lets each year's reduction lower the bill, while a TFP baseline locks last year's higher consumption. Model both before you elect.
You rationalise IPLA by attacking the annual support stream, because the licenses are already bought and the recurring cost is the subscription and support, not a new charge. IPLA programs carry a one time charge at acquisition and a yearly S&S that runs forever unless you act.
The drift is predictable. Db2 tools, OMEGAMON monitors, and utilities accumulate across years and acquisitions, and the S&S keeps billing on entitlements no one maps back to running workload.
Billed rolling four hour average benchmarks this far above sustained workload across the estates we reviewed, before soft capping and offload.
Combined MLC and IPLA saving benchmarks in this band when MSU is optimized first and the contract levers are written in.
The hardware refresh delivers a technology dividend because a newer IBM Z processor does more work per MSU, so the same business workload reports fewer billable MSU on the new machine. The z17, generally available in June 2025, is the current refresh window.
The dividend is real but it is not automatic on every commercial model. On legacy sub capacity AWLC the lower per MSU rates and reduced consumption flow through. On a TFP baseline locked to the old machine, the dividend can be trapped unless you renegotiate the baseline at refresh.
Figure 2. Held flat at a $2,400 per MSU peak baseline, three years cost the same. Optimized first, then refreshed to z17, the effective rate falls toward $1,500 as the technology dividend lands. Benchmark scenario, not a quote.
Seven levers carry most of the value, and each is a clause to write before signing, not a concession to chase later. Each maps to a default in the IBM Z agreement that favors IBM if left untouched.
| Lever | What it protects |
|---|---|
| TFP baseline grandfather | Locks the agreed price per MSU so a future tier or product change prices at today's unit, not a re quote. |
| Growth allowance preservation | Carries the negotiated growth band across the term, so organic growth is pre priced, not surcharged. |
| Sub Capacity alternative protection | A written right to revert to or stay on legacy sub capacity if the workload shrinks, keeping the dividend. |
| IPLA substitution rights | The right to swap retired IPLA tools for current equivalents without a new one time charge. |
| Technology dividend at refresh | A baseline reset at hardware refresh, so a z17 move lowers effective cost rather than freezing it. |
| Data residency posture | Contractual control over where workload and reporting data sit across the mainframe estate. |
| Executive escalation path | A named escalation route and decision window, so the anniversary deadline cannot force a rushed signature. |
The anniversary mechanic underneath all seven is timing. The IBM Z agreement renews on a fixed anniversary, and letting it approach without an optimized baseline hands IBM the clock. Start the MSU program and the negotiation at least 180 days out so the deadline works for you.
The end state is one agreement, sized to optimized MSU, on the right model for your roadmap, with all seven levers written in at signing. That turns a sprawling MLC and IPLA bill into a single negotiation you control.
Sequence is the whole strategy. Optimize MSU, let it settle, set the baseline, then negotiate before the anniversary, so you never argue price on the vendor's clock.
| Line | Vendor shaped path | Optimized |
|---|---|---|
| z/OS MLC (AWLC) | $2,400,000 | $1,800,000 |
| CICS, Db2, IMS, MQ MLC | $2,400,000 | $1,800,000 |
| IPLA support (S&S) | $1,200,000 | $900,000 |
| Annual IBM Z software | $6,000,000 | $4,500,000 |
The arithmetic checks. $1,800,000 plus $1,800,000 plus $900,000 is $4,500,000, against a vendor shaped $6,000,000, so the saving is $1,500,000, a 25 percent cut driven by MSU optimization, baseline discipline, and the technology dividend, not a list discount alone.
Figure 3. The representative IBM Z software stack falls from $6.0M to $4.5M, a $1.5M saving, when MSU is optimized before the baseline and the contract levers are written in. Benchmark scenario, not a quote.
Pull twelve months of SCRT, map MLC and IPLA to running workload, and identify the R4HA headroom on every product.
Set defined and group capacity caps, offload to zIIP, retire idle IPLA tools, and let the lower consumption settle a full quarter.
Set the TFP baseline on the optimized window, write in the seven levers, and sign before the anniversary deadline.
Treat the IBM Z renewal as an MSU optimization problem first and a pricing model problem second. The rolling four hour average peak you carry into the baseline moves more money than the discount on any single line.
Redress Compliance runs this as a buyer side engagement, from the MSU baseline review through the signed agreement. We are glad to tie a meaningful part of the fee to delivered value.
Benchmark ranges: Redress Compliance advisory engagement file, 2024 to 2025.