Start with risk, not fashion
A legacy system can contain years of correct business behavior. Rewriting it from scratch replaces visible technical debt with migration risk. Modernization should begin where change frequency, incidents or performance problems justify intervention.
Typical modernization work
- Map high-risk modules and hidden coupling.
- Move environment-specific configuration out of application code.
- Replace fragile cron or daemon behavior with explicit worker supervision.
- Add database migrations and safer release steps.
- Refactor frequently changed business logic behind clearer boundaries.
- Improve logging and error handling before large structural changes.
Strangler-style replacement when needed
When a subsystem really should be replaced, a gradual boundary can route new behavior through the replacement while the rest of the application continues to run. This keeps migration measurable and reversible.
A rewrite can still be correct
A rewrite is justified when the existing system cannot meet core requirements safely or economically. The decision should follow from constraints and evidence rather than from age alone.