Why Big Bang Rewrites Fail
Over my 30 years in enterprise Java development, I've witnessed countless "big bang" rewrite projects fail spectacularly. The pattern is always the same: leadership decides the legacy system is too complex, approves a multi-year rewrite project, and 18 months later the company has spent millions with nothing to show for it.
The fundamental problem with big-bang rewrites is that they assume you can freeze business requirements while rebuilding the entire system. In reality, your business keeps evolving, your competitors keep moving, and technical debt accumulates in both the old and new systems.
Enter the Strangler Fig Pattern
The Strangler Fig pattern, named after a tree species that grows around its host, offers a better approach. Instead of replacing everything at once, you gradually build new functionality around the legacy system, slowly redirecting traffic until the old system can be safely removed.
How It Works
- Identify a Module: Pick a small, low-risk module in your legacy EJB application
- Build the Wrapper: Create a Spring Boot service that delegates to the existing EJB
- Migrate Gradually: Move business logic piece by piece from EJB to Spring Boot
- Decommission: Once all callers use the new service, remove the old EJB
Real-World Example
One of my clients had a 15-year-old WebLogic application with 200+ EJBs. We started with their user profile service—a simple, well-isolated module with clear boundaries.
Month 1: Created a Spring Boot service that called the existing UserProfileEJB via RMI Month 2: Moved read operations to Spring Data JPA Month 3: Migrated write operations Month 4: Removed the EJB entirely, saving $50K/year in WebLogic licensing
The business never noticed. We deployed weekly. Zero downtime.
Benefits of Strangler Fig
- Risk Mitigation: Each change is small and reversible
- Continuous Value: Deliver improvements incrementally
- Business Continuity: The old system keeps running
- Team Learning: Developers build Spring Boot expertise gradually
When to Use This Pattern
The Strangler Fig pattern works best when:
- Your legacy system is stable enough to wrap
- You can identify clear module boundaries
- You have existing automated tests (or can add them)
- Business requirements allow incremental delivery
It's not ideal when:
- The legacy system is fundamentally broken
- There's extreme technical debt with no tests
- The business demands a complete overhaul immediately
Getting Started
If you're considering modernizing your J2EE application:
- Don't start coding yet. Do an architecture assessment first
- Map your dependencies. Understand which EJBs call what
- Identify your strangling points. Find low-risk modules to start with
- Plan the wrapper layer. Decide how Spring Boot will talk to EJBs (RMI, REST, etc.)
- Start small. Prove the pattern works before tackling complex modules
Conclusion
The Strangler Fig pattern isn't glamorous. It's not a resume-building greenfield project. But it works. I've used it to successfully migrate dozens of legacy J2EE applications to Spring Boot, and I've never had a client go back to the old approach.
If you're maintaining a legacy EJB application and wondering how to modernize without betting the company on a risky rewrite, this pattern is your answer.
Need help planning your strangler migration? Schedule a free consultation to discuss your specific situation.