Plain, sourced papers on the parts of mainframe batch that decide whether a migration works: what sort costs, how to prove the output, where the data traps are, and what the tools and the rules say. Written by Bill Ahlbrandt, who has worked on mainframe sort for more than thirty years.
How IBM bills for z/OS software, why the cost of a sort depends on when it runs, and how to measure it from SMF data before you move it.
What identical output really means when a sort moves off the mainframe: equal keys, EQUALS and NOEQUALS, record counts, return codes, and a test regime that proves it.
The three ways to move batch off the mainframe, what each one does to your sort steps, and what the record says about large rewrites.
What AWS, IBM, Google and Microsoft AI migration tools document about DFSORT and ICETOOL steps, and what translating sort into generated code risks.
Running batch sort on Linux on IBM Z and LinuxONE: how IFL engines are charged, how data moves from z/OS, and when the move pays.
A buyer's guide to replacing DFSORT, Syncsort MFX or CA Sort: invocation routes, compatibility risks, return codes, contracts and a phased plan.
What decides sort throughput on Linux and in the cloud: storage limits, merge passes, memory, containers and the critical path of the nightly batch.
EBCDIC versus ASCII sort order, packed decimal, zoned signs, binary and floating point: where migrated sorts silently go wrong, with worked byte-level examples.
Seventy years of sorting business data, from punched-card sorters and tape merges to DFSORT, Syncsort, sort benchmarks and the z15 sort instruction.
What DORA, the UK operational resilience rules and US third-party guidance mean for moving, or keeping, mainframe batch.
How DFSORT and ICETOOL control statements hide business rules (IFTHEN, OUTFIL, JOINKEYS, SUM), and how to find and score them before a migration.
A unit-cost model for sorting a terabyte nightly on AWS, Azure and Google Cloud, and why architecture matters more than the choice of cloud.
Where sort hides in a z/OS batch estate (JCL, PROCs, COBOL SORT verbs, Db2 utilities) and how SMF records find every one.
What the surveys really say about mainframe skills, why operations knowledge is the scarcer skill in batch, and how to capture it.