Replatforming is a cloud migration strategy where you move an application to AWS and make a limited set of changes on the way, so it runs more cheaply or with less upkeep, without redesigning it. AWS also calls it lift, tinker, and shift or lift and reshape.

It is one of the seven migration strategies (the 7 Rs) in AWS's large migration guidance. AWS describes it as moving the application "and you introduce some level of optimization in order to operate the application efficiently, to reduce costs, or to take advantage of cloud capabilities."

Typical replatforming changes

AWS's own examples include:

  • Moving a Microsoft SQL Server database to Amazon RDS for SQL Server, a managed service, so the team no longer runs the database server itself.
  • Moving virtual machines into containers without code changes (AWS names its App2Container tool for this).
  • Moving to a managed or serverless service in place of a self-managed one.
  • Moving from Windows to Linux, or to AWS Graviton processors.

Where it sits between rehost and refactor

StrategyWhat changesEffortWhen it fits
Rehost (lift and shift)Nothing in the applicationLowestA deadline, such as a data centre exit
ReplatformA few components, often the database or runtimeModerateYou want lower running cost or upkeep without a rewrite
Refactor / re-architectThe architecture, using cloud-native featuresHighestThe application must change to meet business needs

AWS calls refactoring "the most complex and costly of the migration strategies" and does not recommend it for large migrations, because it means modernising during the move. Replatforming is the usual middle ground: AWS notes you "might make a few or many changes", depending on your goals and the target platform.

Practical limits

  • Each change is a change to test. Swapping a database engine version or operating system can break stored procedures, drivers or licensing terms. Plan a test cycle per component, not just per application.
  • Licensing can move the numbers. A move to a managed database or to Linux can change licence costs in either direction. Price it before committing.
  • Keep scope fixed. Replatforming that grows into partial refactoring takes longer and loses the predictability that made it attractive.

For regulated firms, a managed service also shifts some operational work to AWS under the shared responsibility model. The firm still owns configuration, access and data, and needs to record the change for its own outsourcing and resilience records.

For how replatforming compares with the other six strategies, see The 7 Rs of Cloud Migration; for why rehosting alone is rarely the end point, see Lift and Shift Is Not a Strategy. Choosing a strategy per workload is part of a Yemberzal cloud migration assessment.