① Core service · MSU optimization
Under sub-capacity pricing, most IBM Z software is billed on the peak rolling four hour average of MSU consumption. Many ISV contracts price against the same capacity numbers. Engineer the peak down and the entire stack gets cheaper. That is this service.
The peak, not the workload, sets the price.
Monthly license charges for z/OS, CICS, Db2, IMS, and MQ are typically calculated from the highest rolling four hour average (R4HA) your LPARs hit during the month, as captured in the SCRT report you submit to IBM. A batch scheduling collision at 2 a.m. on one Tuesday can price the entire month. The pattern we commonly see: business volume flat, MSU peaks rising, bills following the peaks.
The same capacity figures leak into the rest of the stack. Broadcom (CA) and BMC agreements are typically priced on MIPS or MSU bands, so an unmanaged peak ratchets several vendors' invoices at once. Under Broadcom's consumption programs, measured hourly MSU across production z/OS LPARs feeds the reconciliation directly, a mechanism explained in our True Forward explainer.
IBM's Tailored Fit Pricing replaces the R4HA peak with total consumption at a committed rate. It is the right answer for some estates and an expensive default for others. We treat it as a calculation, never a reflex; the background is in our sub-capacity licensing explainer.
We profile R4HA behavior per LPAR from your SMF data and validate the SCRT reports independently. Most estates discover their peak is driven by a small number of identifiable workloads, and that the SCRT submission itself sometimes overstates what the contract requires you to report.
Know what sets the bill before touching anything.
Peak drivers get named: the batch window collision, the month end report stack, the test LPAR that wandered into production hours. Each driver is classified as movable, cappable, offloadable, or structural, with the MSU value of fixing it quantified.
Every peak has an owner and a price.
Soft capping and defined capacity where service levels allow it, batch resequencing to flatten the window, and zIIP offload for eligible work, since zIIP consumption does not count toward monthly license charges. Specialty engine mechanics are covered in our zIIP, zAAP and IFL explainer.
Lower peaks, same service levels.
With the corrected consumption profile, we model R4HA based sub-capacity pricing against Tailored Fit Pricing and, where relevant, ISV consumption programs. The model decides; whichever path wins on your data is the one we recommend.
The calculation picks the model, not the vendor.
The reduced baseline becomes the price basis at the next renewal: growth bands, caps, and consumption protections written against the corrected numbers, not the inflated history. This is where optimization stops being a one time saving and starts compounding.
Savings that survive the next negotiation.
③ What changes with us in the room
Vendors price against your peaks because nobody challenges the peaks. We challenge the peaks.
Typical reduction negotiated on renewal spend
Mainframe spend negotiated on the buyer side
Engagements delivered since 2019
An MSU, or million service units, measures consumed mainframe capacity. Under monthly license charge pricing, most IBM Z software bills against the peak R4HA of MSU consumption, so one four hour spike can set the month. Many ISV contracts price on the same capacity numbers, which is why one optimization pass typically cuts several publishers' invoices.
No. TFP charges total annual consumption at a committed rate instead of the R4HA peak. Flat, highly utilized estates often benefit; estates with sharp, controllable peaks are frequently better off on sub-capacity pricing with the peaks engineered down. We run the calculation on your data before recommending either path.
Capping and scheduling changes commonly appear in the next monthly SCRT cycle. zIIP offload and pricing model moves typically land over one to two quarters. The full value compounds at renewal, where corrected baselines support the 20 to 35% reductions we typically negotiate.
Usually not at the start. Early wins come from SMF and SCRT data you already produce plus scheduling and capping discipline. Where tooling is justified later, it is evaluated on payback. We sell no software and take nothing from any vendor, so the recommendation is clean.
Related: mainframe cost optimization · renewal advisory · IBM audit defense
Audit notice or renewal under 18 months out? We mobilize within 48 hours.
Get expert help →