MFX: the sort that can cut the MSU that prices it.

Syncsort MFX is a high performance sort, copy, and join utility for z/OS, a drop in alternative to IBM DFSORT that the batch estate leans on for data processing. Now part of Precisely after the Syncsort and Pitney Bowes software lineage came together, it is engineered to do the same sort work in fewer CPU cycles and less elapsed time, and its companion ZPSaver feature offloads sort, copy, and SMS compression cycles to zIIP specialty engines. Because sort is everywhere in batch, MFX is infrastructure: present across most of the estate, depended on by countless jobs, and therefore licensed broadly. Its economics turn on efficiency, which makes it unusual among capacity priced products.

MFX is typically licensed on mainframe capacity measured in MSU, built on the capacity of the LPARs where it is authorized to run. Because it is infrastructure under batch, that footprint usually spans most of the production environment, and often non production too. ZPSaver, which delivers the zIIP offload, is a separately licensable feature, so its entitlement is a distinct line to confirm. Cost tracks capacity, not the number of sort jobs or records, so the controlling questions are whether the licensed MSU still matches the LPARs that genuinely run MFX, and whether the efficiency and offload the product delivers are quantified rather than taken on faith.

Confirm whether ZPSaver is separately entitled and where MFX is actually installed before renewal; both shape the licensed footprint.

The base driver is the licensed MSU of the LPARs running MFX, and because sort is pervasive that footprint tends to be large. The second driver is the ZPSaver feature, a distinct entitlement that buys the zIIP offload. The third is the offset working the other way: MFX is sold on the CPU it saves, so its real cost is the license net of the MSU reduction it delivers across the estate under IBM capacity or consumption charges. That offset is genuine but it has to be measured, because a product that prices on capacity while claiming to reduce it only pays for itself if the reduction is real and captured.

MFX exposure comes from how widely sort spreads and from feature scope. Common traps we see at pattern level:

MFX has a credible substitute in DFSORT, which gives a buyer real room. The five levers that pay:

MFX's principal alternative is IBM DFSORT, which ships with z/OS and is therefore already available at no incremental license. That makes the alternative unusually concrete: the question is not whether a substitute exists but whether MFX's efficiency and zIIP offload save enough CPU to justify paying for it over the bundled sort. Reverting batch from MFX to DFSORT is real work, because job control and exit dependencies accumulate, but it is bounded and well trodden, and the conversion can often be staged. A prepared DFSORT evaluation is one of the more usable walk aways on the mainframe, precisely because the fallback is already installed.