ISO 9001 · ISO 27001 · ISO 20000 certified CERT-In aligned security audits

ETHICSCOMPUTER / JOURNAL · WEB & CUSTOM SOFTWARE DEVELOPMENT · 18 NOV 2026 · 5 MIN READ

Strangler Fig Pattern: Modernising Legacy Software Without a Rewrite

A big-bang rewrite is the riskiest way to replace old software. Learn how the strangler fig pattern replaces a legacy system one module at a time, and how to handle the data migration risks that cause most failures.

IN THIS ARTICLE (8 Sections)

Old software that still runs the business is hard to switch off. It holds years of data and business rules that nobody fully remembers, the original developer may have left, and every change feels risky. The tempting answer is to rewrite everything from scratch and switch over in one weekend. That is also the approach most likely to go wrong.

This guide explains the strangler fig pattern, the step-by-step alternative to a big-bang rewrite, how to choose which module to move first, and the data migration risks that cause most trouble during modernisation, with a checklist for reconciling data before you retire anything.

Quick answer: The strangler fig pattern replaces a legacy system piece by piece. You put a routing layer in front of the old application, rebuild one function at a time in the new system, send traffic for that function to the new code, and retire the old part once it is proven. The business keeps running throughout, and each step can be rolled back.

Why big-bang rewrites go wrong

  • Hidden business rules: discount logic, tax rounding and special cases for one large customer are often buried in code nobody has read for years. A rewrite based on screens and memory misses them.
  • A moving target: the business keeps changing while the rewrite takes months, so the new system is always chasing the old one.
  • No value until the end: users see nothing for a long time, and the budget conversation gets harder every month.
  • One risky cut-over: all data, users and integrations switch at once, and if something breaks there is no easy way back.

How the strangler fig pattern works

The name comes from the strangler fig, a plant that grows around a host tree until it can stand on its own. In software, the new system grows around the old one, function by function, until the old one can be removed. For most legacy software modernization projects, it is the safer default.

  1. Map the current system: modules, database tables, integrations, scheduled jobs and the people who depend on each.
  2. Put a routing layer in front, such as a proxy, API gateway or thin facade, that decides whether the old or new code handles each request. At first, everything still goes to the old system.
  3. Rebuild one function, for example customer statements or order entry, with tests that compare its output against the old system.
  4. Switch traffic for that function to the new code, starting with a small group of users and keeping a way to switch back.
  5. Retire the old code for that function once the new one has run cleanly through a full business cycle.
  6. Repeat until the legacy application has nothing left to do, then decommission it and its server.

Where the old system has no clean entry points, a useful first step is to wrap it with APIs, so new screens and apps read and write data through a stable interface while the internals are replaced behind it.

Choosing which module to move first

The first module sets the tone for the whole programme. Choose one that is low risk but visible enough to build confidence.

Good first candidateWhy it helps
Clear boundaries and few dependenciesLess to untangle and fewer surprises
Read-heavy functions such as reports or statementsMistakes are easy to spot and do not corrupt live data
A real pain point for usersA visible improvement earns support for the next phase
Limited data to migrateThe team practises migration on a small scale first
Not the billing or month-end engineThe most critical process stays stable while the team learns

Data migration risks you must plan for

Code can be rewritten. Data has to survive. In most modernisation projects, the hardest problems come from data, not features.

  • Undocumented fields and codes: a status of 7 or a column called flag2 may drive important behaviour. Find out what every value means before you map it.
  • Dirty data: duplicate customers, items with three spellings, negative stock and records that point to deleted masters.
  • Encoding and format issues: old databases may store text, dates and numbers in formats the new system reads differently, including regional-language text in older encodings.
  • Stored versus calculated values: a balance saved in one table and recalculated in another may not agree. Decide which is the source of truth.
  • Two systems writing the same data: during the transition, old and new parts may both update customers or stock. Give each table one clear owner, or build a tested synchronisation process.
  • History and attachments: decide how many years of history to move, what to archive and how users will reach archived records.
  • The cut-over window: large migrations take time. Rehearse them, time them and schedule them for a low-activity period.
  • Rollback: before you start, know how to switch back and what happens to transactions entered after the switch.

Reconcile before you retire anything

Never retire an old module because the new one looks right. Prove it with numbers:

  1. Compare record counts for every migrated table, by type and status.
  2. Compare control totals: customer balances, stock quantity and value, open order values and tax totals for a period.
  3. Ask key users to check samples they know well, such as their largest customers or most active items.
  4. Run both systems in parallel for at least one full cycle that matters, such as a month-end close, and compare the outputs.
  5. Record every difference, explain it and fix the cause, not just the number.
  6. Get written sign-off from the process owner before switching off the old component.

How EthicsComputer can help

Our legacy software modernization service starts with an assessment of the system, its data and hidden dependencies, often with little documentation left, and gives each module a path: keep, refactor, re-platform or rebuild. Modules are delivered one at a time behind stable interfaces, with data migration scripts, validation reports and a parallel run before anything is retired. If the target is the cloud, our cloud migration team handles the infrastructure, and software architecture consulting can design the target system. Source code and documentation are handed over to your team.

Is your business depending on software nobody wants to touch?

Tell us what the system does and what it runs on, and we will help you plan a safe first step.

About the author

Vinay Grover founded EthicsComputer in Rohtak, Haryana in 2018 and leads its cybersecurity, software and AI work for clients across India and abroad. EthicsComputer is ISO 9001, ISO/IEC 27001 and ISO 20000-1 certified.

Vinay Grover
AUTHOR

Founder & CEO of EthicsComputer (Rohtak, since 2018). Leads the company's VAPT, software architecture and AI automation work. Full profile

END OF ARTICLE

Continue exploring EthicsComputer's technical journal.

EthicsComputer assistant
EthicsComputer Assistant
Automated assistant (AI) • Team replies in office hours
⚡ Fast Response
Instant Quote & Demo
Get a Quote
Verified by SMS
Call Direct
+91 90532 10052
Direct Architect Scoping // Step 1 of 2

Request Fast Quote & Architecture SLA

Receive a preliminary project architecture, pricing tiers, and timeline estimate within 24 hours under strict NDA.

100% Mutual NDA Protected Step 1 of 2 (3 fields)

We use your phone number, company and the details you choose only to reply to your enquiry and prepare a quote. We do not sell your data. Privacy Policy · Your rights