Mizuho and IBM mainframe cost. 71 million dollars out.

Mizuho Financial Group cut roughly 71 million dollars from IBM mainframe licensing through peak engineering, entitlement rationalization, pricing model fit, and a negotiation run last, against a measured estate.

What drove seventy one million dollars of IBM mainframe cost out?

Mizuho Financial Group, one of the largest banking groups in Japan, reduced IBM mainframe related licensing cost by roughly 71 million dollars through a structured optimization of MLC workloads, entitlement baskets, and enterprise agreement terms. The program treated the mainframe bill as an engineering problem first and a negotiation second.

The estate ran core banking on IBM zSeries with the classic stack: z/OS, Db2, CICS, IMS, and middleware, charged through IBM Z software pricing mechanics where monthly license charge tracks the peak rolling four hour average consumption.

Why is the R4HA peak the whole MLC bill?

MLC products bill on the highest rolling four hour average each month, so one badly timed batch window prices the entire month. Workload scheduling, capping, and isolation of dev and test from the billable peak are direct levers on the invoice.

What was in the entitlement basket?

The ELA carried entitlements accumulated over decades, including products no longer deployed and overlapping middleware acquired through bundles. Nobody had reconciled the basket against running workloads in years.

How was the seventy one million delivered?

The program ran four workstreams over multiple quarters: peak engineering, entitlement rationalization, pricing model evaluation, and agreement renegotiation. Each fed the next; the negotiation closed last, against a measured estate.

Sequencing: engineering before negotiation

Every MSU removed from the peak and every dead entitlement dropped from the basket shrank the baseline IBM priced against. Negotiating first would have locked the inflated baseline into a new term; the order of operations was worth a large share of the 71 million.

Where the common advice on IBM mainframe costs is wrong

The standard advice is that mainframe licensing is fixed cost, manageable only through migration off the platform. We disagree. In roughly 8 of the 10 to 15 mainframe programs Morten Andersen has benchmarked, double digit percentage MLC reduction was available within the existing platform through peak engineering and pricing model fit, with no migration risk at all. The buyer side move is to treat the R4HA profile as an engineering artifact you control, and to re fit the pricing model, including tailored fit pricing, to the measured workload before any modernization decision is taken. Migration is a strategy; it is not a license negotiation.

The mainframe bill is not fixed. It is the most engineerable invoice in enterprise software, and the least engineered.

What transfers to other mainframe estates?

The Mizuho pattern transfers to any material zSeries estate: measure the peak, rationalize the basket, fit the model, then negotiate. The scale of saving tracks estate size, but the mechanics are identical at a tenth of the MIPS.

1. Own your R4HA profile: know which workloads set the monthly peak and why.
2. Reconcile entitlements annually: decade old baskets always carry dead weight.
3. Evaluate pricing models on data: the default model is rarely the cheapest for a measured profile.
4. Negotiate last: every unit removed before the table shrinks the priced baseline.

Where does modernization fit?

After the bill is optimized. A rationalized, correctly priced mainframe estate gives modernization programs an honest baseline to beat, and many business cases quietly fail against it.

What to do next

1. Pull twelve months of SCRT reports and identify what sets each monthly peak.
2. Model batch rescheduling and capping against the R4HA profile.
3. Reconcile the entitlement basket against deployed workloads.
4. Evaluate tailored fit and container pricing against the measured profile.
5. Rebuild the ELA negotiation position on the rationalized baseline.
6. Set an annual entitlement and peak review cadence.

The IBM knowledge hub holds the mainframe and ELA guides, and the IBM practice runs programs like this one. More results sit in the case study library.

Frequently asked questions

How did Mizuho save 71 million dollars on IBM mainframe licensing?

Through four sequenced workstreams: R4HA peak engineering, entitlement basket rationalization, pricing model evaluation, and ELA renegotiation run last against the measured, rationalized estate. Engineering before negotiation protected the gains.

What is the R4HA and why does it matter?

The rolling four hour average is the consumption measure IBM monthly license charge bills against, using the highest peak in each month. One badly scheduled batch window can price the entire month, which makes workload scheduling a direct cost lever.

Can mainframe licensing costs really be cut without migrating off the platform?

Yes. In most mainframe programs we have benchmarked, double digit percentage MLC reduction was available within the existing platform from peak engineering, entitlement cleanup, and pricing model fit, with no migration risk.

What is IBM tailored fit pricing?

Tailored fit pricing is an alternative IBM Z software pricing model that replaces R4HA peak billing with consumption based or capacity based structures. Whether it saves money depends entirely on your measured workload profile, so evaluate it on data, not on the pitch.

What should a mainframe estate do before its next ELA renewal?

Pull twelve months of SCRT reports, identify peak setting workloads, reconcile entitlements against deployments, and evaluate alternative pricing models. Take the rebuilt baseline into the negotiation; every unit removed first shrinks what IBM prices.