① Journal · Metrics
Older mainframe contracts were written in MIPS. Modern ones price in MSU. When a vendor converts one to the other at renewal, the conversion factor is a negotiation hiding inside a spreadsheet, and it is where margin quietly moves. Here are three patterns we commonly see at the table, and the checks that keep the conversion honest.
48 hour mobilization Audit notice or renewal under 18 months out? We mobilize within 48 hours.
Get expert help →MIPS is a rough measure of raw processor throughput. MSU, the Millions of Service Units value, is the capacity metric IBM and most mainframe publishers actually license against, derived from the processor model rather than from a clean physics constant. The two do not map by a single fixed ratio. The effective relationship has historically sat somewhere in the range of roughly six to eight MIPS per MSU, but it varies by machine generation and family, and newer processors deliver more MIPS per MSU than older ones. That variability is the whole point. When a vendor migrates a MIPS based contract to an MSU basis, the ratio they choose is not a fact handed down by the hardware, it is a number that lands inside their favor or yours.
Buyers tend to treat the conversion as a technicality and wave it through, which is exactly the reaction the conversion factor is chosen to receive. A small adjustment to the ratio, applied across a large estate, is a material change in the recurring bill that arrives disguised as administrative cleanup.
Across conversions the same three moves recur. None are illegitimate on their face; each shifts value toward the publisher if it goes unexamined:
| Pattern | What happens | The buyer check |
|---|---|---|
| The stale ratio | An old MIPS:MSU ratio is applied to a newer machine that runs leaner | Tie the ratio to the actual installed processor model, not a generic figure |
| The peak anchor | Conversion is based on a historic peak rather than managed consumption | Convert from a cleaned baseline, not the highest number on record |
| The rounded up | MSU figures are rounded upward at every step in the migration | Check rounding direction and the cumulative effect across the estate |
The common thread is that the conversion is presented as settled before the buyer has checked the arithmetic. Each pattern is reversible at the table if it is caught before signature, and effectively permanent once the MSU number becomes the contract of record.
The defense is unglamorous and effective: treat the conversion factor as a negotiated term, not a given. Anchor the ratio to the actual installed hardware. Convert from a managed baseline rather than a historic peak inflated by years of unwatched creep. Audit the rounding. And model the converted MSU number forward, because the figure you accept at conversion becomes the floor every future renewal builds on. A few percent given away in the conversion compounds for as long as the contract lives. This is the same discipline that protects against a bill rising without new workload, covered in MIPS creep, and it sits alongside the broader work of reconciling MIPS and MSU consumption against contracts. When a vendor brings a conversion to the table at renewal, our mainframe license negotiation work makes sure the ratio is examined before it becomes the number you live with. For the capacity metric itself, see the explainer on cost per MSU benchmarks and drivers.