Legacy system modernisation in Malaysia: replace an old system without stopping the business
A legacy system can be modernised without stopping the business by replacing it in phases. A new system takes over one function at a time while the old one keeps running the rest, data is migrated and checked module by module, and the old system is switched off only after the new one has proved itself. The first step is a review of what the old system does.
Macoin TechnologiesSoftware development teamWhen does a system become legacy?
A system is legacy when the business depends on it but cannot change it safely. Typical signs are a platform past its support date, one person who understands the code, no way to connect it to e-Invoice, WhatsApp or a mobile app, and staff copying data into spreadsheets to work around it.
Support dates are the easiest sign to check. Microsoft ended extended support for Windows Server 2012 R2 in October 2023, and its last paid Extended Security Update year ended in October 2026. SQL Server 2014 left extended support in July 2024 and is covered only by paid security updates until July 2027. On php.net, PHP 7.4 reached end of life on 28 November 2022, PHP 8.0 on 26 November 2023 and PHP 8.1 on 31 December 2025, while PHP 8.2 receives security fixes only until 31 December 2026.
Age alone is not the problem. An old system that is supported, documented and does its job may be worth keeping. The case for change comes from risk, such as no security fixes, and from cost, such as hours of manual work or a feature the business needs and cannot get.
Rehost, upgrade, rebuild or replace?
There are four main routes. Rehost moves the system to new servers unchanged. Upgrade brings the platform and code to supported versions. Rebuild writes a new system around the same business rules. Replace swaps it for an off-the-shelf product. The right one depends on how unusual your processes are and how sound the existing code is.
Most real projects mix them. A business might rehost first to get off an unsupported server, upgrade the parts that are sound, rebuild the module that blocks growth, and move a standard function such as payroll to a product.
- RehostFastest and cheapest. Removes the server risk but leaves the code and its limits unchanged.
- UpgradeBrings the platform and libraries to supported versions. Works when the code is reasonably well organised.
- RebuildA new system for processes that give you an edge. Highest effort, and the place for a phased plan.
- ReplaceAn off-the-shelf product for standard functions. Check data export and how your process must change to fit it.
How do you modernise without stopping operations?
Replace the old system one function at a time. A routing layer sends each request to the old or the new system, so users keep working while modules move across. When nothing depends on the old system any more, it is switched off. Microsoft documents this approach as the Strangler Fig pattern.
Microsoft's guidance describes a facade that sits between users and both systems and shifts requests to the new system as each piece of functionality is rebuilt. It warns that the approach does not suit small systems that are simple to replace whole, or cases where requests cannot be intercepted and there is no source code or integration point to work with.
For a business, the practical effect is that each phase has its own go-live, its own testing and its own rollback. Staff learn one new screen at a time, and a problem affects one module instead of the whole company.
How do you move the data safely?
Inventory the data, clean it, map every field to the new system, and run trial migrations before the real one. After each migration, reconcile record counts and totals between the two systems. Keep the old data intact until the new system has been checked, so you can roll back if the numbers do not match.
Data is where modernisation projects slip. Old systems collect duplicate customers, free-text fields used for three purposes and codes that only one person remembers. Finding these during a trial migration costs days. Finding them after go-live costs trust.
Microsoft's example for moving a shared database makes the same point: rollback stays possible while the old tables and synchronisation still exist, and removing the legacy objects should be a deliberate final step, taken only after the new system is validated.
Why does a phased plan reduce risk?
Smaller projects succeed far more often than large ones. In the Standish Group's CHAOS 2020 figures, small projects succeeded about 90% of the time and large projects under 10%. Splitting a replacement into modules turns one large, risky project into a series of small ones, each with a clear finish line.
Phasing also spreads the cost. Each module can be budgeted and approved on its own, and the business sees working results before it commits to the next phase.
How do you start a legacy modernisation project?
Start with a review, not a quotation for a new system. Collect what you can about the old one: server and software versions, source code or the vendor's details, database access, reports staff rely on, and the workarounds people use. The review shows which route fits and which module to move first.
If the original developer has gone, check what you control: the server or hosting account, the domain, the database and any code. A new team can often work from those. Our maintenance and takeover checklist lists the access a new developer needs.
Macoin reviews existing systems before quoting, and we do not promise that every system can be continued. WhatsApp +60 18-263 7350 with what the system does, what it runs on and what is going wrong. Leave out passwords and customer data.
Sources and further reading
Written by

Macoin Technologies
Software development team
Planning guides from Macoin Technologies Holdings Sdn Bhd in Puchong, Selangor. Our public portfolio includes membership apps, a merchant and customer platform, and websites for businesses in Malaysia. Project scope and pricing are confirmed through a quotation.
FAQ
Related questions
It depends on the size of the system and the route chosen. A phased plan shows working results after the first module rather than at the end, and each later phase is estimated after the review.
