the lexicon
The vocabulary
of leaving.
Every field is owned by whoever defines its words. These are the terms of mainframe modernization, legacy modernization, and enterprise computing — defined plainly, kept current, quotable with attribution.
16 terms · the canonical set · cite as “the modernization lexicon, palmdigitalz.com/lexicon”
- Mainframe modernization the discipline
- The discipline of moving the business capability locked in a mainframe estate — its rules, data, and processes — onto modern platforms, without losing correctness, compliance, or continuity. Done well it is evidence-led: strategy, cost, and method decided per application, not by slogan.
- Legacy modernization the broader practice
- The renewal of any aging business-critical system — mainframe or otherwise — by recovering its embedded logic and re-expressing it on current architecture. The mainframe is the hardest case; the same discipline applies down the stack.
- Enterprise computing the territory
- The systems of record that run institutions — core banking, policy administration, settlement, registries — engineered for correctness and continuity above all else. The territory this whole field serves.
- The estate the unit of work
- Everything a legacy platform carries: programs, copybooks, JCL, schedules, data stores, interfaces, and the undocumented decisions of decades. The unit of modernization is the estate, not the application — which is why rationalization comes first.
- Business-rule extraction the foundation
- Recovering the executable business logic — eligibility, pricing, settlement, compliance — from source code into human-readable, verifiable form. The foundation of trustworthy modernization, and the job Palm Ark was built for.
- Source lineage the evidence
- The traceable link from every extracted rule back to the exact lines it derives from. Lineage turns an extraction claim into evidence an auditor can click. A rule without lineage is a rumor with good formatting.
- Near-deterministic extraction the standard
- An extraction pipeline where everything computable is computed and the model answers only checkable questions — producing the same output for the same input, run after run. Three runs, one hash. The framework, in full.
- Application rationalization the triage
- Scoring an estate’s applications by business value, risk, and effort to decide what to modernize, keep, or retire — before any migration is priced. The step most failed programs skipped.
- Dead code the free win
- Code no path calls and no business outcome depends on — typically a third or more of an aged estate, and the cheapest scope reduction in any modernization program. Palm Key finds it before anything is priced.
- The 7 Rs the menu
- The canonical menu of strategies: retain, retire, rehost, replatform, refactor, rearchitect, replace. Most programs fail by choosing one for the whole estate instead of per application. The Exit Doctrine collapses them into the three decisions that matter.
- Rehost — lift-and-shift the half-measure
- Moving workloads to new infrastructure without changing the code. Ends the hardware bill; preserves every line of the legacy problem. A different cell is not freedom.
- Forward engineering the redesign
- Building the target system from extracted business logic rather than translating old code line-by-line — a redesign grounded in recovered rules, not a transliteration. Palm Ray’s whole job.
- Parallel run the proof period
- Operating legacy and target side by side on the same inputs, comparing outputs until the new system has earned production’s trust. The largest hidden line in most cost models.
- Cutover the moment
- The moment production moves to the successor system. In a well-evidenced program it is a decision backed by comparison data — not a leap of faith on a long weekend.
- MIPS · MLC the meter
- The units the mainframe bill is written in: MIPS measure consumed capacity; Monthly License Charges price the software running on it. Together, the recurring cost of staying — model yours here.
- COBOL the language
- The dominant language of enterprise computing since 1959, with hundreds of billions of lines still in production. The language is not the problem — the lost understanding around it is. Can AI convert it?
Fluent now?
Get it in writing.
palm certified · associate tier starts exactly here