Cloud Migration Strategy: Rehost, Replatform, Refactor and More


A cloud migration strategy is a plan that decides, application by application, how each workload will move to the cloud, in what order, and what you will do about risk, cost and downtime.
There is no single best approach. The main options are:
Rehost (lift and shift): move it as is. Fastest, least change.
Replatform: make small changes to gain cloud benefits without rewriting.
Refactor: redesign for cloud, the most work and the most long-term upside.
Repurchase: replace it with a SaaS product.
Retire: switch it off.
Retain: leave it where it is, for now.
Strong strategies mix these approaches. Most enterprise portfolios end up with a little of each, because applications differ in value, age and risk. The skill is choosing the right option for each one.
What Is a Cloud Migration Strategy?
A cloud migration strategy is more than a technical plan. It connects your business goals to the specific actions needed to move systems, data and operations to the cloud.
A good one answers five questions:
Why are we moving? Cost, speed, resilience, data center exit, new capabilities or security?
What are we moving? Which applications, databases and integrations are in scope?
How will each one move? Rehost, replatform, refactor or something else?
In what order? Which first, which last, and what depends on what?
How will we know it worked? Which measures define success?
If you cannot answer the first question clearly, pause. Different goals lead to different strategies. A team racing to exit a data center by a deadline should choose differently from a team trying to cut release times.
For an enterprise-level view of how these decisions fit together, read our guide to cloud migration strategy for enterprises.
Rehost, Replatform, Refactor: The Core Options Compared
Approach | What you do | Effort | Speed | Cloud benefit | Typical risk |
Rehost (lift and shift) | Move the application unchanged to cloud servers | Low | Fastest | Low at first | Carrying inefficiencies and cost into the cloud |
Replatform | Move with targeted tweaks, such as managed databases or containers | Medium | Moderate | Moderate | Hidden dependencies, scope creep |
Refactor (re-architect) | Redesign to use cloud-native services and patterns | High | Slowest | High | Cost, time and complexity |
Repurchase | Replace with a SaaS product | Varies | Moderate | High, if the fit is right | Data migration, process change |
Retire | Decommission it | Very low | Fast | Saves cost | Missed dependencies |
Retain | Keep it on premises for now | None | n/a | None | Postponed decision |
The next sections explain each one in plain terms.
Rehost: Lift and Shift
Lift and shift means moving an application to cloud infrastructure with little or no change. Think of it as moving house with all your furniture in the same arrangement.
When it fits:
You have a hard deadline, such as a data center contract ending.
The application is stable and not worth redesigning.
You want a quick first win and plan to improve later.
You lack the skills or time for deeper change.
What to watch:
Cost surprises. Applications built for fixed servers can be expensive on pay-as-you-go cloud pricing if they are oversized or always on.
Missed benefits. You get the location of the cloud without elasticity, automation or managed services.
"Temporary" becoming permanent. Many lift-and-shift workloads never get revisited.
Lift and shift is a legitimate choice, not a failure. It becomes a mistake only when it is chosen by default, without a plan for what comes next.
Replatform: Lift, Tinker and Shift
Replatforming keeps the core application but swaps certain parts for cloud services. Common examples are moving a self-managed database to a managed one, adopting containers, or using managed load balancing and storage.
When it fits:
You want real operational savings without a rewrite.
Parts of the stack are costly to maintain, such as databases, patching or backups.
The application is valuable but not changing much.
What to watch:
Scope creep. "Small tweaks" can grow into a redesign. Define the boundary up front.
Compatibility. Managed services behave slightly differently from self-run ones, so test carefully.
For many applications, replatforming is the best balance of effort and return.
Refactor: Re-Architect for the Cloud
Refactoring means redesigning the application to take full advantage of cloud capabilities, such as microservices, serverless functions, event-driven design and automated scaling.
When it fits:
The application is strategic and needs to scale, change fast or enter new markets.
The current design blocks growth, reliability or release speed.
You have the budget, skills and time for a longer project.
What to watch:
Cost and duration. This is the most expensive approach and the slowest to pay back.
Architecture fashion. Not every application needs microservices. Read our guide to monolith vs microservices before assuming it does.
Parallel running. You may run old and new systems side by side for a while.
If refactoring is on the table, our posts on legacy modernization from monolith to microservices and serverless cloud solutions for enterprises cover the patterns in depth. This work falls under what many teams call application modernization.
Repurchase, Retire and Retain
These three options are often overlooked, and they can be the cheapest wins.
Repurchase. Some applications are better replaced than moved, such as an old in-house CRM replaced with a SaaS platform. This avoids carrying custom code forward. For help weighing the trade-offs, see our build vs buy framework that accounts for technical debt.
Retire. Most application portfolios contain systems nobody uses, or whose function is duplicated elsewhere. Switching them off removes cost, risk and migration effort. Always confirm dependencies first.
Retain. Some workloads should stay put for now: they may be due for replacement, tied to specialized hardware, or limited by regulation or latency. Retaining by choice is fine. Retaining because nobody decided is not. If you are weighing where certain workloads belong, our comparison of bare metal vs cloud infrastructure helps.
How to Choose the Right Approach for Each Application
Score every application on a few simple dimensions, then match it to an option.
Question | If the answer is... | Lean toward |
Is it still needed? | No or rarely | Retire |
Is there a good SaaS replacement? | Yes | Repurchase |
Is it strategic and constrained by its design? | Yes | Refactor |
Is it valuable but costly to run? | Yes | Replatform |
Is there a hard deadline and low appetite for change? | Yes | Rehost |
Are there regulatory, latency or hardware blockers? | Yes | Retain, for now |
Then refine with these factors:
Business value: how important is the application to revenue or operations?
Technical health: how old, fragile or well documented is it?
Dependencies: what does it connect to, and what depends on it?
Compliance: what data does it hold, and which rules apply? Our guides to HIPAA technical controls and SOC 2 compliance show what auditors look for.
Team skills: can your people run it in the cloud?
Cloud Migration Strategy Steps for Enterprises
Here is a practical sequence enterprises follow. The exact order varies, but each stage matters.
Step 1: Define goals and success measures. Write down the business outcomes and how you will measure them. Examples are lower run cost, faster releases, better resilience or a data center exit date.
Step 2: Take inventory. List every application, database, integration and owner. Discovery tools help, but also talk to the people who run the systems, because documentation is often out of date.
Step 3: Map dependencies. Find out what talks to what. Hidden dependencies cause most migration surprises and unplanned downtime.
Step 4: Assess and categorize. Assign each application one of the options above, using business value, technical health and constraints.
Step 5: Design the landing zone. Build the secure foundation first: accounts, networking, identity, access controls, logging and cost controls. Networking deserves special care, and our Cloud Networking Solutions team covers design, integration and automation.
Step 6: Build a migration plan and waves. Group applications into waves, starting with lower-risk workloads to build skill and confidence. Save the most critical and tangled systems for later.
Step 7: Migrate data carefully. Data is often the hardest part: volume, downtime windows and integrity checks all matter. See our guide to data migration to the cloud.
Step 8: Test thoroughly. Validate function, performance, security and failover before cutover. Our guide to cloud and modernization testing lists what to verify.
Step 9: Cut over with a rollback plan. Use staged releases where possible, and always define how you would return to the previous state.
Step 10: Optimize and operate. Migration is the start, not the finish. Tune cost and performance, train teams and set up ongoing monitoring and support.
Cloud Migration Checklist
Use this quick list before each wave. For a longer walkthrough, see our cloud migration checklist with 7 steps to a smooth transition.
Before migration
Business goals and success measures agreed
Application inventory complete, with owners
Dependencies mapped
Migration option chosen for each application
Landing zone, identity and network design approved
Security and compliance requirements documented
Budget and cost controls defined
Rollback and communication plans ready
During migration
Data migrated and verified for completeness
Performance and security tested against targets
Integrations reconnected and validated
Cutover window agreed with business owners
Monitoring and alerts live before traffic moves
After migration
Old environment decommissioned safely, not left running
Cost reviewed against forecast
Performance tuned and resources right-sized
Runbooks and on-call coverage in place
Lessons recorded for the next wave
Security, Reliability and Cost: Do Not Treat Them as Afterthoughts
Three areas decide whether a migration is judged a success.
Security. Responsibilities change in the cloud. Build security into pipelines and infrastructure from the start instead of adding it later. Our guide to DevSecOps implementation shows how, and our DevSecOps Consulting team can assess your posture.
Reliability. Moving to the cloud does not make systems reliable by itself. Plan for failure, monitoring and recovery. See what reliability engineering in the cloud involves.
Cost. Cloud costs scale with usage, so waste scales too. Tag resources, set budgets, right-size regularly and review often. If you run containers, our post on Kubernetes optimization covers practical savings, and our guide to managing Kubernetes across multiple clusters applies once you grow.
Common Cloud Migration Mistakes
Starting without clear goals. "Move to the cloud" is not a goal.
Treating all applications the same. One strategy for everything wastes money or time.
Skipping discovery. Unknown dependencies break things on cutover day.
Lift and shift with no follow-up. You pay cloud prices for data center design.
Underestimating data migration. It often takes longer than the applications.
Ignoring skills. Teams need training, or support from people who have done it before.
Forgetting governance. Without cost and access rules, sprawl starts immediately.
Declaring victory at cutover. Operations and optimization are where the value is realized.
If you want an outside view before committing money, our guide to what technology advisory involves explains how that works.
Frequently Asked Questions
What is a cloud migration strategy?
A cloud migration strategy is a plan that decides how each application and data set will move to the cloud, in what order, and how risk, cost, security and downtime will be managed. It connects business goals to specific technical actions.
What is the difference between rehost, replatform and refactor?
Rehost moves an application unchanged to the cloud. Replatform makes targeted changes, such as using a managed database, without redesigning the application. Refactor redesigns the application to use cloud-native architecture. Effort and long-term benefit rise from rehost to refactor.
What is lift and shift in cloud migration?
Lift and shift is another name for rehosting. The application is moved to cloud infrastructure with little or no change. It is fast and low risk to execute, but it may not deliver cost or performance benefits without later optimization.
Which cloud migration approach is best?
No single approach is best. The right choice depends on each application's business value, technical condition, dependencies, compliance needs and your timeline. Most enterprises use a mix.
What are the main steps in a cloud migration strategy?
Define goals, take inventory, map dependencies, categorize applications, build the landing zone, plan migration waves, migrate data, test, cut over with a rollback plan, then optimize and operate.
How long does a cloud migration take?
It depends on portfolio size, complexity, data volume and the approaches chosen. Rehosting a few simple workloads can be quick. A full enterprise migration with refactoring is typically a multi-phase program.
What are cloud migration services?
Cloud migration services help organizations plan and carry out the move, including assessment, strategy, landing zone design, data and application migration, testing, security and ongoing operations.
What does cloud migration cost?
Cost depends on scope, approach and how much change you make. Rehosting is usually cheapest to execute, refactoring the most expensive. Also account for data transfer, duplicate running costs, testing, training and ongoing cloud spend.
Start With Your Application List
You do not need a finished plan to get useful advice. You need a list.
Send us your application inventory, even if it is a rough spreadsheet with names, owners and a guess at business importance. Our cloud architects will sort each entry into rehost, replatform, refactor, repurchase, retire or retain, flag the dependencies most likely to cause trouble, and suggest a first wave that is safe to start with. You get a triage you can react to, not a slide deck of generic advice.
Begin the conversation through our IT Advisory & Consulting team. When the plan moves into delivery, our DevOps & Continuous Delivery engineers automate the pipelines, Quality Engineering validates each wave before cutover, and Managed Support Services runs the environment afterward. If your applications also need redesign, Smart Product Engineering and Platform Engineering cover the modernization work.
Author Name - Mrunalini Wankhede