Architect interviews shift from “what does PERFORM do?” to “how should three hundred payroll programs share FILE layouts across a decade?” You may defend keeping Easytrieve in the portfolio, design a migration wave to Broadcom 11.6 function mode, or explain how batch windows absorb Black Friday volume without LPAR storage incidents. Panels include enterprise architects, mainframe infrastructure leads, and application directors who care about risk, cost, audit, and team velocity. This page collects architect-level questions with model answers emphasizing governance, patterns, coexistence with COBOL and ETL, and measurable outcomes. Speak in tradeoffs, not absolutes: the best architects explain why the organization chose a path and what they would revisit next quarter.
Architect answer: charter a working group with reps from each application area and operations; baseline against Broadcom manuals; publish naming, formatting, error handling, compile PROC, and promotion checklist; provide pattern library (validation framework, report shell macros); enforce via Endevor approver rules and automated compile gates where possible; quarterly review for release changes. Success metric: smaller review diffs, fewer macro-related incidents, faster onboarding. Avoid standards nobody reads—keep under fifteen pages with links to internal tutorials.
Single owner team, semantic versioning or parallel macro names for breaking changes, mandatory regression compile list per macro tier, LIST output stored in change ticket, separation of dev/test/prod libraries, audit trail in change control. Breaking positional parameters without version bump is a governance failure—architects design contract tests comparing expansion snapshots.
| Artifact | Purpose |
|---|---|
| Coding standards manual | Consistent review and audit |
| Macro regression matrix | Safe library promotion |
| Pattern catalog | Reuse validation and report shells |
| Compile PROC tiers | Dev DEBUG vs prod optimized |
| Runbook template | Ops handoff for new jobs |
| Migration playbook | Release upgrade waves |
Mandate patterns for high-risk repeats: inbound feeds (validation framework), corporate listings (report framework), file copy utilities (standard COPY pattern). Allow ad hoc for low-volume one-offs with sunset dates. Architect judgment: pattern tax (learning curve) versus incident cost. Reference section 29 pattern pages as shared vocabulary in design reviews.
Typical layers: extract (Easytrieve or ETL), validate/transform (Easytrieve), report (Easytrieve REPORT), handoff to GL or warehouse (Easytrieve or COBOL). Architects define dataset contracts between layers—record layout macros are the API. Version macro when contract changes; consumers recompile. Discuss idempotent reruns and GDG conventions.
Architect response framework: inventory programs by change frequency, skill pool, integration needs, and batch window cost. Retain Easytrieve where report maintenance is stable and team skilled. Consider migration for programs needing modern API, scarce Easytrieve skills, or heavy integration with cloud lakes. Hybrid is normal—rip-and-replace rarely passes ROI unless regulatory or vendor pressure. Offer phased strangler: new feeds in new tech, stable reports remain until business event forces move.
Division of labor: DFSORT presorts and merges at scale; IDCAMS defines VSAM; Easytrieve consumes ordered files and produces reports; COBOL handles complex transaction modules or CICS. Architect defines where sort lives to avoid duplicate sort tax. Integration via shared copybooks/macros for record layouts where shops align COBOL and Easytrieve definitions.
Phases: automated scan reserved words and deprecated syntax; tier by criticality; pilot application; macro library first; compatibility compiles with warning dashboard; fix waves by team capacity; parallel run totals for financial tiers; function-mode enforcement date; hypercare with rollback modules. Architect adds communication plan, training, and executive metrics—percent converted, defect rate, window impact.
Impact assessment, budget request, roadmap slot, risk register entry, alternative evaluation, stakeholder alignment. Show you track Broadcom release notes proactively—not react after compile failures in production.
Dependency graph, critical path analysis, resource pools (VFM, SORT WORK, tape), throttle non-critical jobs, pre-allocate temp space, capacity test at 120% volume annually, incident playbooks. Easytrieve-specific: identify VIRTUAL-heavy and high-BUFNO jobs for review before peak. Architects partner with operations—not throw code over the wall.
Macro library promotion without regression; DEBUG left in production; storage below 16 MB from buffer sprawl; silent validation skips; GDG wrong generation; undocumented restart requirements. Mitigate with standards, checklists, and monitoring SYSPRINT anomaly detection where automated.
Facilitate decision matrix: change frequency, parameter needs, tooling support, skill base. Document ADR (architecture decision record). Pick one default for new work; grandfather legacy until touch. Architects decide and explain—not endless debate.
Tutorial path, pattern library, mentor pairing, safe sandbox LPAR, code review rubric, gamified quiz pages, rotate through ops shadowing. Measure time-to-first-solo promotion as success.
The architect interview is planning the whole city's bus routes—not driving one bus. You decide where stations go, how buses share roads, and what happens when a road closes. You still know how a bus engine works so nobody sells you a route map that ignores rivers.
1. Enterprise FILE layouts should be owned by:
2. Easytrieve versus COBOL at architect level:
3. Function-mode migration strategy needs:
4. Design patterns in Easytrieve programs help architects:
5. Batch window architecture considers: