A practical guide for IT and engineering leaders who need to decide, application by application, how to move to the cloud.
In short
A cloud migration strategy assigns each application one of a small set of options, known as the Rs: retire, retain, rehost, replatform, repurchase or refactor, with AWS now adding relocate as a seventh. Choose per application, based on its business value, technical health and the skills you have, and modernise after moving unless the migration itself creates the opportunity.
A cloud migration (molnmigrering) is rarely one decision. A typical mid-sized company runs dozens of applications, and the right move for the finance system is not the right move for an old internal tool or a customer-facing product. This guide is written by Devies, a Gothenburg-based product and engineering consultancy, for CIOs, CTOs and IT managers planning a move to AWS, Azure or another cloud. It explains the Rs, gives a step-by-step method for choosing between them, and lists the mistakes that make migrations slow and expensive.
What is a cloud migration strategy?
A cloud migration strategy is the approach chosen for moving a specific workload to the cloud, from switching it off to rebuilding it completely. The strategies are usually called the "Rs", and the classic list has six of them.
AWS Prescriptive Guidance now lists seven strategies, adding relocate to the original six. Microsoft's Cloud Adoption Framework uses a similar list with slightly different names, splitting refactor into refactor, rearchitect and rebuild. The ideas are the same.
Strategy
What it means
When it fits
Retire
What it means
Switch off or archive the application
When it fits
No business value left, or it duplicates another system
Retain
What it means
Keep it where it is for now
When it fits
Recently upgraded, data residency rules, hardware dependencies, or not worth moving yet
Rehost ("lift and shift")
What it means
Move it unchanged to cloud virtual machines
When it fits
Stable application, short deadline, no near-term need to change it
Relocate (AWS only)
What it means
Move a whole virtualised platform as it is
When it fits
Large VMware-style estates that need to move quickly
Replatform ("lift and reshape")
What it means
Move with small changes, such as a managed database or containers
When it fits
Clear savings from managed services with little code change
Repurchase ("drop and shop")
What it means
Replace it with a SaaS product
When it fits
Commodity functions such as CRM, HR or collaboration
Refactor or rearchitect
What it means
Redesign the application to use cloud-native capabilities
When it fits
The application holds back the business and must change anyway
Rehost is the fastest and least risky to execute, but it carries every existing problem into the new environment. Refactor delivers the most long-term value, but it is also the most complex and costly. Most of the work in a cloud migration strategy is deciding which applications deserve which treatment.
How to choose a cloud migration strategy, step by step
The reliable way to choose is to inventory every application, tie each to a business driver, rule out the obvious cases first and then decide the rest one by one. These eight steps work for a handful of applications as well as a large portfolio.
- Build the inventory. List every application with its owner, users, dependencies, data, licences, hosting and support status. Discovery tools help, but owners must confirm what the tools find.
- Name the business goals. Write down why you are migrating: closing a data centre, reducing cost, improving resilience, enabling AI and data work, or speeding up delivery. Each goal favours different strategies.
- Retire first. Remove applications with little or no use before planning anything else. AWS suggests looking for applications with very low utilisation or no inbound connections over the last 90 days as candidates.
- Mark what must be retained. Note applications tied to special hardware, strict data residency or a pending vendor SaaS release.
- Find the SaaS candidates. For commodity functions, compare a SaaS replacement with moving the current system.
- Assign a strategy to the rest. For each remaining application, choose rehost, replatform or refactor based on its technical health, how long it will live and how much it needs to change. Validate each choice with the business owner.
- Group into waves. Move applications in waves that respect dependencies, and start with a low-risk pilot wave to prove the landing zone, security, networking, monitoring and cost controls.
- Optimise after moving. Review cost and performance after each wave, right-size resources, and plan modernisation of the applications that need it.
Write the decision and its reason next to each application. When priorities change, you can revisit the reasoning instead of starting over.
Which mistakes make cloud migrations expensive?
Most expensive cloud migrations fail on planning rather than technology: they move problems unchanged, try to modernise everything at once, or skip the operating model. Five mistakes recur.
- Rehosting problem applications. Microsoft's guidance is blunt: rehosting does not resolve existing performance, reliability or architectural issues, and carries technical debt forward. It also advises rehosting only when you are confident the workload will not need modernising within two years.
- Refactoring everything during the move. AWS does not recommend refactoring in large migrations, because modernising while migrating is the most complex option. Move first, then modernise, unless there is a clear reason not to.
- Expecting automatic savings. Servers moved unchanged and left running around the clock often cost more in the cloud. Savings come from right-sizing, managed services and switching off what is not used.
- Ignoring the operating model. Cloud changes how teams deploy, monitor, secure and pay for systems. Without infrastructure as code, cost ownership and new routines, the old problems return.
- Deciding everything up front. Organisations that are new to the cloud learn a lot in the first wave. Leave room to change the plan.
Devies saw the value of moving gradually at Inet, where a CTO-as-a-Service engagement planned a step-by-step move towards DevOps in the cloud, starting with selected parts of the system. According to the case, this meant Inet did not have to decide on cloud, on-premises or hybrid right away, and automation and infrastructure as code became key criteria for technology choices.
When should you modernise during a cloud migration?
Modernise during the migration when the application must change anyway, when your team has the skills and time, or when the migration budget is the realistic chance to fund it. Otherwise, move first and modernise afterwards.
The Cloud Adoption Framework gives three conditions for modernising during migration: the team has the required skills and time, the workload needs compatibility updates such as replacing unsupported components, and the migration unlocks funding and stakeholder support that might not return later.
For large or business-critical applications, modernisation rarely has to be a big-bang rewrite. The strangler fig pattern lets you move one capability at a time into the cloud while the old system keeps running. Before splitting an application into many services, check whether a well-structured single application would serve you better; our comparison of microservices and a monolith helps with that call. And if the real problem is years of shortcuts in the code, read our guide to technical debt as a business risk first.
Devies
Want a second opinion?
Devies, a Gothenburg-based product and engineering consultancy, works on questions like this with Nordic product and IT teams. Tell us where you are and we will suggest a sensible first step.
Frequently asked questions
What are the 6 Rs of cloud migration?
The 6 Rs are six strategies for moving an application to the cloud: retire (switch it off), retain (keep it where it is), rehost (move it unchanged), replatform (move it with small optimisations), repurchase (replace it with SaaS) and refactor (redesign it for the cloud). AWS now lists a seventh, relocate, for moving whole virtualised platforms. Each application in a portfolio gets its own strategy.
What is the difference between rehost and replatform?
Rehosting, often called lift and shift, moves an application to cloud virtual machines without changing it. Replatforming moves it with a few targeted changes to benefit from cloud services, such as moving a database to a managed database service or packaging the application in containers. Replatforming takes a little longer but usually reduces operating effort and cost compared with a pure rehost.
How long does a cloud migration take?
It depends on the number of applications, their dependencies and the strategies chosen. Rehosting a single stable application can take weeks, while a portfolio of dozens of systems usually moves in waves over many months. Refactoring a business-critical system can take longer still. A pilot wave early in the programme gives you real data to plan the rest.
Is cloud migration cheaper than running our own servers?
Not automatically. Moving servers unchanged and running them continuously can cost more than an existing data centre. Savings come from retiring unused applications, right-sizing resources, using managed services, switching off non-production environments when they are not used and owning costs per team. Many companies migrate for resilience, security and speed of delivery as much as for cost.










