Most legacy system modernization initiatives don’t fail as a result of the know-how was incorrect. They fail as a result of management selected the riskiest doable solution to deploy it: changing a whole system in a single cutover occasion and hoping nothing breaks.
The information on this isn’t ambiguous. Impartial analysis from NRI Digital Consulting places the failure or underperformance charge of modernization initiatives at 68 to 79 p.c in 2026. A separate business evaluation discovered that over 50 p.c of enterprise IT transformation packages run over finances and greater than 30 p.c are considerably delayed. When TSB Financial institution tried a single-event “huge bang” migration of 5.2 million buyer accounts to a brand new core banking platform in April 2018, the consequence was 1.9 million clients locked out of their accounts, widespread fraud publicity, a GBP 330 million loss, a GBP 48.65 million regulatory superb, and the resignation of the CEO. Knight Capital misplaced 440 million {dollars} in 45 minutes in 2012 when a single deployment script did not replace one among eight manufacturing servers, reactivating dormant legacy code that had been sitting untouched for practically a decade.
Neither of those failures occurred as a result of modernization was the incorrect name. They occurred as a result of the organizations tried to switch deeply embedded legacy methods in a single irreversible transfer, and not using a mechanism to isolate threat, take a look at incrementally, or roll again safely.
The strangler fig sample exists particularly to forestall this consequence. It’s the modernization technique that permits you to exchange legacy software logic module by module, maintain the enterprise working all through, and keep away from betting the corporate on a single go-live weekend. This text breaks down precisely the way it works, when it applies, the place it fails, and how you can construct the enterprise case for utilizing it.
What Is the Strangler Fig Sample?
The strangler fig sample is an incremental legacy system modernization technique by which a brand new system is constructed and deployed alongside an current legacy system, progressively taking on performance piece by piece till the legacy system may be absolutely decommissioned.
The time period was coined by Martin Fowler in 2004 after observing strangler fig vines within the rainforests of Queensland, Australia. A strangler fig germinates within the cover of a number tree, grows downward and outward, and slowly envelops the host over years. Ultimately the host tree dies and decomposes, leaving the fig standing in a kind that echoes the unique tree’s form. Fowler noticed the identical sample in software program: as an alternative of demolishing a legacy system and rebuilding from scratch, you develop a brand new system round it till the outdated one is now not wanted.
Technically, the sample works by means of an interception layer, often a reverse proxy, API gateway, or facade, that sits between customers (or consumer purposes) and the underlying methods. Initially, this layer routes one hundred pc of visitors to the legacy system, precisely as earlier than. As new parts are constructed, examined, and validated, the routing layer incrementally redirects particular requests to the brand new system whereas the whole lot else continues to stream to the legacy software. Over time, increasingly more performance shifts to the brand new system till the legacy system handles nothing in any respect and may be safely retired.
That is basically completely different from a rewrite. A rewrite treats modernization as a single challenge with a single supply date. The strangler fig sample treats modernization as a steady, reversible course of with dozens of small supply factors as an alternative of 1 massive one.
Why Large Bang Legacy System Substitute Fails So Usually
Enterprises maintain selecting the massive bang method as a result of it appears sooner and cleaner on a challenge plan. The truth documented throughout enterprise IT packages tells a distinct story.
The numbers are constant throughout unbiased research.
NRI Digital Consulting’s 2026 evaluation discovered that 68 to 79 p.c of modernization initiatives fail or underperform, pushed primarily by weak stakeholder alignment, incomplete system evaluation, and ineffective challenge administration, not by the underlying know-how selection. A separate 2026 business evaluation discovered that failed modernization packages sometimes eat 18 to 36 months of organizational power earlier than being deserted or scaled again, and that the enterprise is commonly left with a legacy system that has now fallen even additional behind as a result of no function improvement occurred throughout the failed try.
Underestimating legacy enterprise logic is the primary root trigger.
Legacy methods accumulate many years of undocumented enterprise guidelines, exceptions, and workarounds that had been by no means written down anyplace besides within the code itself. When a workforce commits to a full rewrite, they’re implicitly betting that they will reconstruct all of that embedded logic appropriately, in a single try, earlier than go-live. The TSB Financial institution autopsy, performed by the regulation agency Slaughter and Could at TSB’s personal request, discovered that the financial institution’s board was advised about solely 800 of the two,000 identified defects current on the time of go-live, and that the 2 knowledge facilities constructed to assist the brand new platform had been, regardless of being specified identically, configured inconsistently. These are the sorts of gaps that floor in an enormous bang cutover with no fallback path.
The federal authorities supplies the starkest illustration of this at scale.
A U.S. Authorities Accountability Workplace report discovered that the federal authorities spends greater than 100 billion {dollars} yearly on IT, with roughly 80 p.c of that consumed by working and sustaining current methods slightly than constructing new functionality. Of 10 vital legacy methods flagged for modernization again in 2019, solely three had been absolutely modernized by 2025. A number of of the remaining methods nonetheless run on COBOL and Meeting, languages with a shrinking pool of engineers who can safely keep them.
Self-discipline in completion is a separate failure mode solely.
Even organizations that select an incremental method can stall out as soon as the highest-visibility options are migrated, leaving the legacy system working indefinitely in a diminished however everlasting capability. This creates a worst-case consequence: double the upkeep burden, as a result of each the outdated and new methods now require ongoing assist, with not one of the payoff of full decommissioning.
How the Strangler Fig Sample Works: The Technical Mechanics
Implementing the strangler fig sample for software modernization follows a constant sequence, no matter whether or not you’re modernizing a monolithic net software, a mainframe COBOL system, or a legacy ERP platform.
The first step: map system boundaries.
Earlier than writing any new code, the legacy system must be understood as a set of discrete parts slightly than a single indivisible block. This discovery section is the place most timeline overruns truly originate. Enterprises that skip detailed dependency mapping constantly see a six-month challenge increase right into a multi-year engagement as soon as testing surfaces features no one knew existed. This step alone determines whether or not the remainder of the challenge succeeds.
Step two: outline skinny, independently deployable slices.
Break the system into parts sufficiently small emigrate individually, however important sufficient to ship actual enterprise worth on their very own. A reporting module, an authentication service, and an invoicing engine are typical beginning slices as a result of they have an inclination to have clearer boundaries and decrease interdependency than core transaction processing.
Step three: introduce the routing layer (the facade).
That is the software program seam that makes your entire sample doable. For net purposes, that is often a reverse proxy or API gateway. For legacy mainframe or desktop methods, it usually requires a custom-built wrapper or adapter layer. In the beginning, the facade routes all visitors to the legacy system, so nothing adjustments from the consumer’s perspective.
Step 4: construct and validate the brand new element in isolation.
The workforce builds the trendy alternative for one slice, ideally beginning with a element that already has robust take a look at protection and relatively low technical debt. AWS’s personal prescriptive steering on the sample recommends this explicitly: begin with the element that provides the workforce probably the most confidence, not the one which delivers the flashiest headline.
Step 5: incrementally redirect visitors.
As soon as the brand new element is validated, the facade begins routing a portion of related requests to it whereas the legacy system continues dealing with the whole lot else. This may be carried out progressively, by proportion of visitors, by consumer section, or by function flag, permitting the workforce to observe for regressions earlier than committing absolutely.
Step six: repeat, then decommission.
This cycle of constructing, redirecting, and validating repeats slice by slice. As soon as the legacy system has zero reside visitors routed to it, it may be decommissioned. Some organizations select to maintain the facade completely as an adapter layer for older shoppers whereas persevering with to evolve the core system behind it. Others take away it solely as soon as migration is full.
All through this whole course of, the legacy system stays absolutely operational. That single reality is what eliminates the highest-risk failure mode within the TSB and Knight Capital instances: an irreversible cutover with no secure path backward.
Strangler Fig Sample vs. Large Bang Rewrite
When the Strangler Fig Sample Is the Proper Name, and When It Isn’t
The strangler fig sample isn’t a common reply. It suits greatest when particular circumstances are current.
Use it when:
- The legacy system is massive, business-critical, and can’t tolerate prolonged downtime
- The system has identifiable, separable modules (authentication, billing, reporting, and related features)
- The group must maintain delivery options and can’t take up a multi-year function freeze
- Regulatory or compliance necessities (DORA, HIPAA, PCI-DSS) demand steady auditability slightly than a disruptive cutover
- Management needs measurable progress and ROI checkpoints slightly than a single binary consequence 18 months out
Rethink it when:
- The legacy system is sufficiently small {that a} full rewrite carries restricted threat
- The structure is so tightly coupled that no significant element boundaries exist, which means “skinny slices” can’t be outlined with out deep refactoring first
- The group lacks the operational self-discipline to keep up a routing layer and dual-system knowledge synchronization for an prolonged interval
- Management is unwilling to decide to ending the migration, since a stalled strangler fig effort leaves each methods working indefinitely at double the upkeep price
The Enterprise Case: What Modernization Finished This Means Really Delivers
C-suite sponsors don’t want a metaphor about rainforest vines. They want numbers.
McKinsey’s 2026 analysis on enterprise modernization packages accomplished in 2024 and 2025 reveals infrastructure price reductions of 30 to 50 p.c, together with a 20 to 30 p.c enchancment in improvement cycle instances following legacy-to-cloud migration. A separate synthesis of the identical physique of labor discovered launch cycles accelerating 40 to 60 p.c and safety breach threat dropping by roughly half after modernization.
The price of inaction is equally nicely documented. Enterprises presently spend 60 to 80 p.c of complete IT finances merely sustaining current methods, based on Gartner and PwC analysis, and most organizations undercount their true legacy price burden by 40 to 60 p.c as a result of bills are scattered throughout engineering time, end-of-life vendor contracts, integration workarounds, and unbudgeted safety remediation. Gartner individually discovered that organizations nonetheless working legacy methods are 40 p.c extra more likely to fail compliance audits below frameworks like GDPR, HIPAA, and DORA.
There may be additionally a expertise dimension that boards more and more must deal with as a threat line merchandise, not an HR footnote. Reuters reporting has documented that a big share of banking methods nonetheless run on COBOL, and the inhabitants of engineers who can keep that code is retiring sooner than it’s being changed. That is exactly the type of threat the strangler fig sample is constructed to handle: it permits a corporation emigrate away from a shrinking-talent-pool language module by module, slightly than requiring a single, high-stakes rewrite executed below time stress by a dwindling variety of specialists.
Frequent Pitfalls to Plan For Earlier than You Begin
- Information synchronization complexity: Working two methods in parallel means each usually want entry to the identical underlying knowledge. With out a clear synchronization technique, knowledge drift between the legacy and new system turns into a silent supply of manufacturing incidents. Plan this earlier than writing any migration code, not after the primary slice is already in manufacturing.
- Dropping the self-discipline to complete: The only most cited failure mode particular to this sample is stopping as soon as probably the most seen, highest-value parts have been migrated. Set a tough decommissioning goal for the legacy system in the beginning of this system, not as an afterthought as soon as momentum slows.
- Treating it as a purely technical train: Legacy methods are sometimes inflexible as a result of the organizational construction and decision-making processes round them had been by no means modified. Migrating the code with out addressing workforce possession, determination rights, and cross-team coordination reproduces the identical rigidity within the new system. Conway’s Legislation applies right here immediately: the software program will mirror the communication construction of the group that builds it.
- Skipping acceptance standards for enterprise logic: Legacy methods usually encode undocumented guidelines that solely floor in manufacturing edge instances. Outline acceptance standards for correctness and completeness earlier than migration begins, and run the identical take a look at suite in opposition to each the outdated and new system to verify nothing was silently dropped.
AI-Native Legacy Modernization Companies with the Strangler Fig Sample
Legacy modernization succeeds whenever you perceive the system earlier than altering it. ISHIR makes use of AI-assisted code evaluation to uncover hidden enterprise logic, dependencies, and modernization alternatives throughout COBOL, Java monoliths, and legacy ERP platforms, lowering discovery time and migration threat.
We mix the Strangler Fig Sample with Generative AI to modernize purposes incrementally. New companies are constructed, examined, and deployed in phases whereas the legacy system stays operational, enabling sooner supply with minimal enterprise disruption.
Whether or not you’re modernizing a legacy software, ERP, or monolithic platform, ISHIR delivers a phased, AI-powered modernization roadmap with measurable outcomes. The result’s decrease threat, sooner execution, and a future-ready structure designed for steady innovation.
Is Your Legacy Software Slowing Innovation and Rising Modernization Danger?
Modernize confidently with ISHIR’s AI-powered legacy modernization method utilizing the Strangler Fig Sample to cut back threat, speed up migration, and maintain your online business working with out disruption.
FAQs
Q. What’s the strangler fig sample in software program modernization?
The strangler fig sample is an incremental method to legacy system modernization the place a brand new system is constructed alongside an current one, progressively taking on performance by means of a routing layer till the legacy system may be absolutely decommissioned, with out ever taking the enterprise offline.
Q. Why is the strangler fig sample higher than an enormous bang migration?
It distributes threat throughout many small, reversible adjustments as an alternative of concentrating it in a single cutover occasion. If a brand new element underperforms, the routing layer can redirect visitors again to the legacy system instantly, one thing that’s successfully unimaginable as soon as a full rewrite has gone reside and the outdated system has been decommissioned.
Q. How lengthy does a strangler fig migration sometimes take?
Timelines fluctuate by system complexity and organizational self-discipline, however organizations that prioritize their highest-impact methods first sometimes attain constructive ROI inside 12 to 14 months, based on business knowledge on accomplished 2024 to 2025 modernization packages. Full rewrites, by comparability, usually take 36 to 48 months earlier than producing a return and carry a considerably increased failure charge.
Q. Does the strangler fig sample work for mainframe modernization?
Sure. It is among the most typical approaches for mainframe modernization particularly as a result of mainframe methods sometimes run mission-critical, zero-downtime workloads. Kyndryl’s 2025 analysis discovered that organizations modernizing purposes on the mainframe utilizing incremental approaches noticed a median ROI of 288 p.c.
Q. What are the largest dangers of utilizing the strangler fig sample?
The 2 most vital dangers are knowledge synchronization complexity between the legacy and new methods working in parallel, and organizational self-discipline: stalling the migration after the highest-value parts are carried out, which leaves each methods working indefinitely and doubles ongoing upkeep prices.
Q. Can AI speed up a strangler fig modernization program?
Sure. Trade knowledge from 2026 reveals AI-assisted modernization, together with automated dependency mapping, code translation, and take a look at era, lowering modernization challenge timelines by 40 to 50 p.c in comparison with 2023 ranges. This is applicable on to the strangler fig sample’s discovery and slice-definition phases, which have traditionally been probably the most time-consuming components of the method.
Q. Who must be liable for the routing layer in a strangler fig migration?
The facade or routing layer wants a single accountable proprietor, sometimes a platform or structure workforce, as a result of it turns into vital infrastructure at some point of the migration. Microsoft’s Azure Structure Heart particularly warns in opposition to letting the facade develop into a single level of failure or efficiency bottleneck, which suggests it wants the identical monitoring and reliability requirements as any manufacturing system.
