Application Migration

Last reviewed: August 2026

Overview

This is the work of moving applications and infrastructure from on-premises or another cloud into the cloud. It’s not simply copying VMs — it’s a process of re-evaluating the entire application’s architecture, dependencies, and operational model.

The 7R migration strategy

The 7R framework, proposed by Gartner and extended by AWS, is the standard for classifying migration strategies per workload.

Strategy	Description	Difficulty	Effect
Retire	Discontinue a workload that’s no longer needed	Low	Immediate cost savings
Retain	Keep it on-premises (hybrid)	Low	No migration cost
Relocate	Move as-is at the hypervisor level (e.g., VMware Cloud on AWS)	Low	Fast transition with minimal change
Rehost	VM-level Lift & Shift	Low	Fast transition, limited cloud benefits
Replatform	Partial transition to managed services (e.g., DB to RDS)	Medium	Reduced operational burden
Repurchase	Switch to SaaS (e.g., a homegrown CRM → Salesforce)	Medium	Delegated operational responsibility
Refactor	Redesign as cloud-native (serverless, microservices)	High	Maximizes scalability/efficiency

Strategy selection criteria

Situation	Recommended strategy
Tight data center exit deadline	Relocate or Rehost
Low business value, short-lived workload	Retire
Functionality already available as a commercial solution	Repurchase (SaaS)
Core workload where you want to maximize cloud benefits (scalability, cost)	Refactor
Legacy system that needs stable operation with no changes	Retain
Want to reduce operational burden while keeping the code	Replatform

Lift-and-shift vs. refactoring trade-offs

A comparison of the two most common options: Rehost (Lift & Shift) and Refactor.

Item	Rehost (Lift & Shift)	Refactor (cloud-native)
Migration duration	Weeks to months	Months to years
Development cost	Low (minimal change)	High (redesign/reimplementation)
Operational cost	Similar to on-premises	Can be reduced via cloud optimization
Scalability	Limited (VM-based)	High (serverless, horizontal scaling)
Disaster recovery	Retains existing approach	Cloud-native HA/DR
Risk	Low	High (redesign may fail)
Leveraging cloud benefits	Limited	Maximal
Data center exit deadline	Can meet even short deadlines	Requires a long timeframe

Migration process

Large-scale migration is a project that goes through multiple phases.

Phase	Key activities
1. Discovery	Collect inventory — identify servers, applications, dependencies, and usage
2. Assessment	Decide the 7R strategy per workload, estimate cost, analyze risk
3. Planning	Decide migration order (waves), rollback plan, downtime budget
4. Landing Zone	Build the cloud account/network/security foundation (see Landing Zone)
5. Migration	Perform the actual data/application transfer
6. Validation	Performance/functional testing, user acceptance testing
7. Cutover	Traffic switchover, heightened monitoring
8. Optimize	Optimize the cloud environment (cost, performance, security)

Migration waves

Migrating hundreds or thousands of workloads at once is risky. Migration is typically carried out in waves.

Pilot wave — Build experience with simple, low-risk workloads (e.g., internal tools, dev environments)
Core waves — Group applications and migrate them sequentially
Critical wave — Migrate business-critical workloads last (after sufficient validation)

Minimizing downtime

Production workloads need to minimize downtime.

Technique	Description	When to use
Continuous block-level replication	Continuously replicates changed blocks from the source VM to the target	The default approach for most VM migration tools
Blue/Green cutover	Run old and new infrastructure in parallel, then switch via DNS/LB	Web services, APIs
Database CDC	Continuously replicate DB changes so cutover completes within minutes	DB migration
Phased cutover	Gradual transition by user/region/feature	Services with large user bases

Cutover checklist

Items to verify at the actual cutover moment.

Migration tools

Assessment and discovery

Vendor	Product	Capability
AWS	Application Discovery Service	Agent-based/agentless on-premises inventory collection
AWS	Migration Hub	Central migration dashboard
Azure	Azure Migrate	Assessment plus integrated server/DB migration
Google Cloud	Migration Center	Portfolio assessment, dependency mapping
OCI	Cloud Migrations	Integrated assessment and execution

VM/server migration

Vendor	Product	Capability
AWS	Application Migration Service (MGN)	Block-level replication. Minimal-downtime Rehost
Azure	Azure Migrate: Server Migration	VMware/Hyper-V/physical servers → Azure VM
Google Cloud	Migrate to Virtual Machines	VMware/AWS/Azure → Compute Engine
OCI	OCI Cloud Migrations	VMware/AWS → OCI

Containerization

Vendor	Product	Capability
AWS	App2Container	Containerize Java/.NET apps
Azure	Migrate to containers	ASP.NET/Java → AKS
Google Cloud	Migrate to Containers	VM → GKE containers

Common mistakes

Starting migration without Discovery/Assessment — Without understanding dependencies and traffic patterns, failures occur post-cutover and rollback becomes necessary.
Migrating workloads before a landing zone exists — Moving without account structure, networking, IAM, and a tagging scheme in place means everything has to be rebuilt later.
Trying to Refactor all workloads at once — Schedules collapse and quality issues arise. For most, Rehost followed by gradual improvement is more realistic.