① Guide · Cost optimization
Development and test capacity should not pay production rates, yet on most estates it quietly does. Non production workload is one of the most overpaid lines on the frame. Separate it, price it on the right vehicle, and write the boundary into the contract.
Non production load at production prices. It adds up fast.
On a typical estate, a meaningful slice of licensed capacity is not production at all. It is development, test, QA, training, and disaster recovery rehearsal. Because that workload often runs inside the same sub-capacity scope as production, or on LPARs that roll into the same capacity based agreements, it gets billed at the production rate. The software vendors are happy to leave it there, because non production capacity bought at production prices is pure margin.
The opportunity is that non production workload is the easiest to move and the easiest to isolate. It does not always need production scale, it rarely needs production data, and it can often be carved onto separate LPARs or taken off the machine entirely. Each of those is a lever, and each gives you a credible threat the vendor has to price against. The work is to measure the non production footprint, then choose the right mechanism for each piece.
| Lever | How it works | Best fit |
|---|---|---|
| Move off the machine (zD&T) | IBM Z Development and Test Environment runs genuine z/OS on x86 by emulation; eligible dev and test leaves the licensed footprint entirely | Dev and test that does not need production scale or production data |
| Isolate non production LPARs | Separate dev, test, and training onto distinct LPARs so the capacity can be scoped and priced apart from production | Workload that must stay on the frame but can be partitioned |
| New application incentives (zNALC) | IBM New Application License Charge offers a reduced MLC rate for qualifying new z/OS workloads in their own LPAR | Greenfield applications being built and tested before go live |
| Vendor contract carve outs | Negotiate a defined, discounted non production tier with each software vendor, documented so it cannot drift back to production pricing | Third party tools where dev and test entitlement is bundled at full rate |
| Sub-capacity scope discipline | Ensure non production work is not inflating the four hour rolling average that drives sub-capacity MLC | Estates where dev and test runs in the production sub-capacity scope |
Directional and pattern level. Eligibility, qualifying conditions, and product terms for zD&T, zNALC, and sub-capacity reporting change and vary by contract, so confirm the current rules and your own agreement terms before modeling a move. Note: zD&T, zNALC, and LPAR are vendor terms and keep their official spelling.
Quantify how much licensed capacity is dev, test, QA, training, and DR rehearsal rather than production. Until you can size it, the vendor controls the framing. The measurement is the evidence the whole negotiation rests on.
Some workload moves off the machine onto zD&T, some isolates onto separate LPARs, some qualifies for a new application charge, and some only needs a contract carve out. Map each non production component to the mechanism that prices it lowest.
The leverage is that you can isolate or relocate the workload. Faced with a discounted non production tier or losing that capacity from the licensed base, vendors commonly find a number. The threat only works if the plan behind it is real.
Define the non production scope precisely and document it, so the discounted tier is enforceable and cannot quietly revert to production pricing at the next renewal. An undocumented carve out is a discount with an expiry date you did not agree to.
Pay production rates for production. Nothing else.
Separating non production workload and pricing it on the right vehicle removes one of the quietest overpayments on the mainframe. The saving compounds, because non production capacity tends to grow with every project and, left unscoped, every increment is billed at the production rate. Designed correctly, the carve out holds across renewals instead of resetting.
Match the work to a service with cost optimization or, when it lands at renewal, renewal advisory. For the mechanics, the soft capping explainer and contractual versus consumed MSU show how sub-capacity scope drives the bill. For the IBM context see the IBM buyer side guide, and where dev and test sits inside a Broadcom bundle, the Broadcom 18 month plan folds it into the renewal.
Usually not. Non production workload is one of the most overpaid lines on the frame. zD&T, isolated LPARs, new application charges, and contract carve outs all price it lower, but the discount has to be designed in.
IBM Z Development and Test Environment runs genuine z/OS on x86 by emulation, moving eligible dev and test off the production machine so that capacity no longer counts toward sub-capacity or third party capacity licensing.
Measure the non production footprint, then take it to each vendor as an evidenced request. The leverage is that you can isolate or relocate the workload, so the vendor faces a discounted tier or losing the capacity entirely.
Only if the boundary is written into the contract. Document the non production scope precisely so the tier is enforceable and cannot drift back to production pricing at renewal.
Audit notice or renewal under 18 months out? We mobilize within 48 hours.