Case Study 1: GlobalBank Five-Year Modernization Roadmap
The Context
GlobalBank's board asked a simple question: "Should we replace our COBOL systems?" Priya Kapoor's answer was a 30-page analysis comparing three options.
The Three Options
Option A: Full Replacement
- Convert all 1.2 million lines of COBOL to Java
- Migrate from z/OS to cloud-native architecture
- Estimated cost: $850 million over 10 years
- Risk: 40-60% failure rate based on industry data
- Key concern: Business rule migration accuracy
Option B: Status Quo
- Continue maintaining COBOL as-is
- No new tooling or processes
- Estimated cost: $8 million/year (current run rate)
- Risk: Talent gap as senior developers retire; no integration with modern channels
- Key concern: Inability to serve mobile and digital customers
Option C: Modernization (Selected)
- Keep COBOL core, modernize everything around it
- CI/CD, APIs, hybrid architecture, AI tools, talent development
- Estimated cost: $4.2 million over 5 years
- Risk: Medium (incremental approach reduces risk)
- Key concern: Requires sustained organizational commitment
The Decision
The board chose Option C based on three factors: 1. Economics: $4.2M vs $850M — a 200:1 cost ratio 2. Risk: Incremental modernization has no single point of failure 3. Speed: Year 1 deliverables were already partially complete (CI/CD, initial APIs)
Year-by-Year Plan
Year 1 (Foundation): Git migration, CI/CD pipeline, z/OS Connect, knowledge transfer program. Budget: $800K.
Year 2 (Integration): 20 COBOL APIs exposed, event streaming, ADDI deployment, apprenticeship program. Budget: $900K.
Year 3 (Analytics): Cloud data warehouse via CDC, analytics dashboards, AI documentation project, 3 new junior hires. Budget: $950K.
Year 4 (Optimization): Refactor 50 high-complexity programs, 80% test coverage, selective cloud batch migration. Budget: $850K.
Year 5 (Maturity): Full hybrid architecture, API catalog of 100+ services, team at 8 developers, continuous modernization process. Budget: $700K.
Projected Outcomes
| Metric | Before | After (Year 5) |
|---|---|---|
| Deployment frequency | 2/month | 12+/month |
| API endpoints | 0 | 100+ |
| Test coverage | ~5% | 80% |
| Mobile/web integration | Screen scraping | Native REST APIs |
| Mean time to recover | 4 hours | 30 minutes |
| Developer team size | 5 | 8 |
| Average developer age | 46 | 38 |
Discussion Questions
- What assumptions must hold true for Option C to succeed? What could go wrong?
- If the talent gap worsens (no new hires available), how would you modify the plan?
- The board approved the plan in 22 minutes. What does this suggest about the quality of the presentation?
- At what point, if ever, should GlobalBank reconsider Option A (full replacement)?