Skip to main content
Legia

IT Infrastructure & Cloud

Rehost, Replatform, or Refactor: Cloud Migration Strategies

Not every system moves to the cloud the same way. Rehost, replatform, and refactor offer different balances of speed, risk, and savings, and some systems are better left where they are.

SEO and Content Specialist

5 min read

Cable connectors on a network panel

Short answer: rehost (lift and shift) moves a system to the cloud without changing its structure: the fastest and lowest-risk option, but the bill is often not yet lower. Replatform swaps some components for cloud services, such as a managed database, so the bill follows usage and part of the maintenance shifts to the provider. Refactor redesigns the system for the cloud: the most expensive and slowest, suitable only when the system really has to change. Some systems are better left in place.

The decision to move systems to the cloud is often made at the company level (“we are moving to the cloud”), although the execution has to be decided system by system. An old accounting application, a company website, a file server, and an internal application built in-house all have different needs. The most widely used framework for sorting them is the migration “Rs”, which AWS guidance breaks down into seven strategies. This article covers the three most often used, plus the option of not moving at all.

Comparing three migration paths and one stay-put path

AspectRehost (as is)Replatform (reorganized)Stay in place
What is doneMoved to a cloud server without changing the structureSome components replaced with cloud servicesLeft on the current server, with its connection to the cloud set up
Time requiredShortestLonger, with adjustments and retestingNo move
Change riskLow, the way the system works does not changeMedium, run in parallel for longerNone, as long as the server is still fit for use
The bill afterwardOften not yet lower, sized after the old serverFollows actual usageThe cost of hardware and your own maintenance keeps running
MaintenanceThe operating system is still managed by youPartly shifts to the providerEntirely in your team’s hands

Refactor is not in the table because it is more like a new development project than a move: the application architecture is redesigned to use cloud services fully.

Illustration: cloud migration

Rehost: move first, optimize later

Rehost suits cases where the old server must be left soon, for example when the data center contract ends or the hardware is no longer supported, or when the system is hard to change because no one understands the code anymore. AWS guidance also recommends this pattern for large-scale migrations: move first, then modernize once the system is running in the cloud.

What to watch for: a cloud server sized to mimic the old one is often too large for actual usage. Without tuning after the move, the monthly bill can be higher than the cost of the old server.

Replatform: small changes, real benefits

Replatform replaces the components that are hardest to maintain with managed services. The most common example is moving a database from your own server to a managed database service, so that backups, updates, and monitoring are handled by the provider. The application code changes only slightly.

Because there are changes, the old and new systems run in parallel for longer before the switch, and testing is more thorough.

Refactor: only when the system really has to change

Refactor redesigns the application to take full advantage of cloud capabilities, for example by splitting a large application into several services. It has the highest cost and the longest timeline. It makes sense when the old application is already holding the business back, for example because it is hard to develop, hard to test, or cannot handle growth. For many companies, refactoring is better scheduled after the migration is complete, not done at the same time.

Staying in place is also a valid decision

Not every system needs to move. A system with stable usage, hardware that is still fit for use, or a dependency on special on-site equipment can stay on its current server, with a secure connection to other cloud services. A system that is no longer used is better retired than moved along with the rest.

Common mistakes during migration

  • Moving everything at once. One problem along the way can stop the entire operation. A phased move leaves room to learn from the first system.
  • Not preparing a rollback plan. If the system in the cloud has problems after the switch, it must be clear how to return to the old system and how long that takes.
  • Forgetting dependencies between systems. A migrated application can slow down because its database is still on the old server, separated by network distance.
  • Assuming the cloud is automatically more secure. Access rights, backups, and monitoring still have to be set up; a misconfiguration in the cloud is as dangerous as on your own server.

How to choose per system

  1. List all systems along with their owners, usage, and dependencies on other systems.
  2. Mark the urgent ones, for example servers whose hardware support is about to end.
  3. Assess how easy each is to change. Systems without documentation or without anyone who understands them tend to be rehosted.
  4. Calculate the bill after the move, not only the cost of the move itself. A comparison of the cost of your own server against the cloud is in the article Office Server, Cloud, or Shared Hosting.
  5. Sequence the moves so that the riskiest system is not moved first.

How Legia does it

In the Cloud Migration & Architecture service, a readiness assessment sorts each system into one of the paths above, then sets the order of moves and an estimate of the monthly bill. Moves are done in phases and run in parallel with the old system, with a rollback plan if needed. Legia handles AWS, Google Cloud, Microsoft Azure, and local providers. For mid-sized workloads, the whole process usually takes 2 to 4 months, ending with resource labels, budget limits, and usage alerts in place. Other infrastructure services are on the IT Infrastructure & Cloud page.

Frequently asked questions

Is rehost always the wrong move because it does not save money?

No. Rehost is often the right first step, especially when time is short. Savings come from right-sizing and replatforming after the system is running in the cloud.

Can one company use all three strategies at once?

Yes, and that is the most common case. The strategy is chosen per system, not per company.

How long are the old and new systems run in parallel?

Until the new system has proven stable. The length depends on the change risk: rehost is usually short, replatform longer.

Sources

Does your office server or network need to be set up and looked after? See Legia IT infrastructure and cloud services, or schedule a consultation through the contact page.

Category IT Infrastructure & Cloud

Keep reading

Related articles

All Articles