This chapter covers cloud migration strategies — the different approaches you can take when moving on-premises workloads to Azure. Understanding these strategies is critical for the AZ-900 exam because they form the foundation of cloud adoption decisions and appear in questions about migration planning and total cost of ownership. Cloud Concepts (Objective 1.1) carries approximately 25-35% of the exam weight, and migration strategies are a key subtopic. By the end of this chapter, you'll be able to distinguish the five main strategies (rehost, refactor, rearchitect, rebuild, and replace) and identify which scenario each fits.
Jump to a section
Imagine your company is moving to a new office building. You have three options. Option 1: 'Rehost' — you literally pick up your desks, computers, and filing cabinets and move them as-is to the new building. It's fast but you bring all the old dust and tangled cables. Option 2: 'Refactor' — you take your existing furniture, disassemble it, and reassemble it in a more modern layout. You keep the same pieces but improve the arrangement. Option 3: 'Rebuild' — you sell all your old furniture and buy brand-new, space-saving furniture designed for the new building. This takes the most time but gives you the best long-term efficiency. In cloud migration, 'rehost' (lift-and-shift) moves applications unchanged to Azure IaaS. 'Refactor' (also called rearchitect) modifies applications to use PaaS services like Azure SQL Database instead of running your own SQL Server. 'Rebuild' (rearchitect) rewrites applications from scratch using cloud-native technologies like Azure Functions and Cosmos DB. The mechanism: each strategy moves different layers of the stack (OS, middleware, data, code) with varying degrees of change. Rehost moves everything at the VM level; refactor changes the data and app layers; rebuild changes all layers. The cost and time trade-off mirrors the moving analogy: rehost is fastest but may not reduce operational overhead; rebuild is slowest but maximizes cloud benefits like auto-scaling and pay-per-use.
What Are Cloud Migration Strategies?
Cloud migration strategies are the different approaches organizations use to move their on-premises IT resources — applications, data, and infrastructure — to the cloud. The business problem they solve is that not all applications are equally suited for the cloud. Some are old, monolithic, or tightly coupled to specific hardware. Others are modern, modular, and ready for cloud-native features. Choosing the right strategy saves time, money, and reduces risk.
Microsoft defines five primary strategies, often remembered by the acronym '5 Rs': Rehost, Refactor, Rearchitect, Rebuild, and Replace. Each strategy represents a different level of change to the application and infrastructure. The exam expects you to know the definition of each and when to apply it.
How Migration Strategies Work — Step by Step
The migration process typically follows these steps:
Discovery and Assessment — Use tools like Azure Migrate to inventory on-premises servers, databases, and applications. Assess dependencies and performance metrics.
Strategy Selection — For each workload, choose one of the 5 Rs based on business goals, technical constraints, and cost.
Migration Execution — Move the workload using Azure services like Azure Site Recovery, Azure Database Migration Service, or manual redeployment.
Optimization — After migration, right-size resources, implement auto-scaling, and apply Azure cost management.
The Five Migration Strategies (5 Rs)
#### 1. Rehost (Lift-and-Shift) - What it is: Moving applications to Azure with minimal changes. You migrate VMs as-is using Azure Migrate or Azure Site Recovery. - When to use: Quick wins, urgent datacenter exit, or when applications are not easily modified. - Azure services: Azure Virtual Machines, Azure Migrate, Azure Site Recovery. - Pros: Fast, low risk of breaking functionality, no code changes. - Cons: No cloud optimization (still pay for OS licenses, no auto-scaling), operational overhead remains.
#### 2. Refactor (Lift and Shift + Optimize) - What it is: Making small changes to applications to take advantage of PaaS services without rewriting the entire codebase. - When to use: When you want to reduce operational costs and improve scalability without a full rewrite. - Azure services: Azure App Service, Azure SQL Database, Azure Kubernetes Service (AKS). - Example: Move a web app from running on IIS VMs to Azure App Service, and replace on-premises SQL Server with Azure SQL Database. - Pros: Reduced management overhead, built-in high availability, auto-scaling. - Cons: Requires some development effort, may need minor code changes.
#### 3. Rearchitect - What it is: Modifying the application architecture to be more cloud-native, often by breaking a monolithic app into microservices. - When to use: When you need to improve scalability, resilience, or deployment speed significantly. - Azure services: Azure Service Fabric, Azure Kubernetes Service, Azure Functions, API Management. - Example: Split a monolithic e-commerce app into separate microservices for inventory, orders, and payments, each running in containers. - Pros: High scalability, fault isolation, independent deployment. - Cons: Significant development effort, longer timeline, requires skilled teams.
#### 4. Rebuild - What it is: Rewriting the application from scratch using cloud-native technologies. - When to use: When the existing application is outdated, unsupported, or cannot meet new business requirements. - Azure services: Azure Functions, Cosmos DB, Azure DevOps, Azure Logic Apps. - Example: Replace a legacy CRM with a new application built on Azure Functions and Cosmos DB. - Pros: Maximum cloud benefits, modern architecture, lower long-term maintenance. - Cons: Highest cost and time, business disruption during development.
#### 5. Replace - What it is: Instead of migrating the existing application, replace it with a SaaS (Software as a Service) solution. - When to use: When a SaaS alternative meets business needs and reduces operational overhead. - Azure services: Dynamics 365, Office 365, or third-party SaaS. - Example: Replace an on-premises email server with Exchange Online. - Pros: No migration effort for the application itself, predictable subscription pricing. - Cons: Loss of customization, data sovereignty concerns, vendor lock-in.
Key Components and Tools
Azure Migrate: Central hub for discovery, assessment, and migration. It provides dependency visualization, cost estimates, and integration with migration tools.
Azure Site Recovery: Orchestrates replication and failover of VMs for rehost migrations.
Azure Database Migration Service: Migrates databases (SQL Server, MySQL, PostgreSQL) to Azure with minimal downtime.
Azure App Service Migration Assistant: Assesses and migrates web apps to App Service.
Pricing and TCO Considerations
Migration strategies directly affect total cost of ownership (TCO). Rehost may have lower upfront costs but higher ongoing operational expenses (OS licenses, management). Refactor and rearchitect reduce operational costs by using PaaS services, which include automatic patching, backups, and scaling. Rebuild has the highest initial cost but the lowest long-term operational overhead. Replace converts capital expenditure (licenses, hardware) into operational expenditure (subscription).
Comparison to On-Premises
On-premises, you own everything: hardware, software, facilities. Migration strategies essentially decide how much of that ownership you transfer to Azure. Rehost transfers the infrastructure but leaves the application unchanged. Refactor transfers the platform. Rebuild transfers the entire stack to cloud-native services. Replace outsources the application entirely to a SaaS provider.
Azure Portal and CLI Touchpoints
In the Azure portal, migration tools are found under 'Azure Migrate' service. You can create a project, discover servers, assess readiness, and start migration. For example, to replicate a VM using Azure Site Recovery via CLI:
az site-recovery vault create --name myVault --resource-group myRG --location eastus
az site-recovery replication-protection-container create --fabric-name myFabric --protection-container-name myContainer --vault-name myVault --resource-group myRGFor database migration:
az dms create --name myDMS --resource-group myRG --location eastus --sku Standard_1v2
az dms project create --name myProject --service-name myDMS --resource-group myRG --source-platform SQL --target-platform SQLDBDiscover and Assess On-Premises Environment
Use Azure Migrate to discover your on-premises servers, applications, and dependencies. Install the Azure Migrate appliance (a lightweight VM) that collects performance data, software inventory, and network dependencies. This step identifies which workloads are suitable for each migration strategy. Azure Migrate also provides cost estimates by mapping on-premises resources to Azure VM sizes. For the exam, remember that discovery is the first step in any migration project and that Azure Migrate is the primary tool.
Choose a Migration Strategy per Workload
For each workload, evaluate business requirements (uptime, performance, cost) and technical constraints (compatibility, licensing). Select one of the 5 Rs. For example, a legacy accounting app with no source code might only be suitable for rehost. A modern web app with a SQL Server backend could be refactored to use Azure App Service and Azure SQL Database. Document the chosen strategy and expected benefits.
Migrate Using Appropriate Azure Tools
Execute the migration using the tool that matches your strategy. For rehost, use Azure Site Recovery to replicate VMs to Azure and perform a failover. For refactor, use Azure Database Migration Service for databases and deploy application code to App Service via deployment slots. For rebuild, develop a new application using Azure DevOps pipelines and deploy to Azure Functions or AKS. Each tool handles data synchronization and cutover with minimal downtime.
Validate and Optimize Post-Migration
After migration, test the workload thoroughly to ensure functionality, performance, and security. Use Azure Monitor and Application Insights to track metrics. Right-size resources (e.g., change VM size, enable auto-scaling). Implement Azure Cost Management to set budgets and alerts. For refactored or rebuilt apps, enable automated backups, geo-replication, and disaster recovery. This step is often overlooked but is critical for realizing the benefits of migration.
Decommission On-Premises Resources
Once the workload is stable in Azure and all data has been synchronized, decommission the on-premises hardware and software. This includes turning off servers, removing network connections, and cancelling maintenance contracts. Proper decommissioning avoids double costs and security risks. Azure Migrate can help track the status of migrated resources. For the exam, know that decommissioning is the final step in the migration lifecycle.
Scenario 1: Data Center Exit for a Retail Company A mid-sized retailer needs to close its on-premises data center within six months due to expiring lease. They have 200 VMs running a mix of Windows and Linux servers, including a custom inventory management system and a SQL Server database. The IT team chooses rehost for the inventory system (no source code available) and refactor for the SQL Server database (migrate to Azure SQL Database Managed Instance). They use Azure Migrate for assessment and Azure Site Recovery for VM replication. The migration is completed in four months, but the inventory system still requires manual patching and has no auto-scaling, leading to higher operational costs than anticipated. The database refactor reduces management overhead by 40%.
Scenario 2: Modernizing a Legacy Banking Application A bank has a monolithic loan processing application built on .NET Framework 4.5 and SQL Server 2012. The application is critical but difficult to scale. The bank decides to rearchitect by breaking the monolith into microservices (using Azure Kubernetes Service) and moving the database to Azure Cosmos DB for global distribution. The project takes 18 months and costs $2M, but the new system handles 10x the transaction volume and reduces downtime from monthly to near-zero. The key lesson: rearchitecting is expensive but necessary for performance-critical applications.
Scenario 3: Replacing an On-Premises Email System A non-profit organization uses an on-premises Exchange Server 2010, which is out of support. They have limited IT staff and budget. They choose the replace strategy: they migrate all mailboxes to Exchange Online (Office 365). Using the Exchange admin center and migration batches, they move 500 mailboxes over a weekend. The migration is seamless, and the organization saves $30,000 per year in hardware and licensing costs. However, they lose some custom email retention policies, requiring a policy review. Common mistake: underestimating the time to train users on the new interface.
What AZ-900 Tests on This Objective (1.1) The exam tests your ability to distinguish the five migration strategies and identify which strategy is appropriate for a given scenario. You will not be asked to perform migration steps; rather, you'll see scenario-based multiple-choice questions like: 'A company wants to move a legacy application to Azure with minimal changes. Which migration strategy should they use?' (Answer: Rehost).
Most Common Wrong Answers and Why Candidates Choose Them 1. Confusing Rehost with Refactor: Candidates often pick 'Refactor' when the question says 'minimal changes.' Refactor implies changes; Rehost implies none. Trap: If the question mentions 'optimize for cloud benefits,' then Refactor is correct. 2. Choosing Rebuild for everything: Rebuild sounds like the best option, but it's the most expensive and time-consuming. The exam wants you to recognize that Rebuild is only for specific cases (e.g., unsupported technology). 3. Mixing up Rearchitect and Rebuild: Rearchitect modifies the architecture (e.g., microservices) but reuses existing code. Rebuild starts from scratch. The exam may describe a scenario where the app is 'redesigned using cloud-native services' — that's Rebuild, not Rearchitect. 4. Forgetting Replace as a strategy: Many candidates only remember four Rs. Replace is often the correct answer when a SaaS alternative exists (e.g., replacing Exchange with Office 365).
Specific Terms and Values That Appear Verbatim - 'Rehost' is also called 'lift-and-shift.' - 'Refactor' is sometimes called 'replatform.' - 'Rearchitect' focuses on changing the application architecture. - 'Rebuild' means rewriting the application. - 'Replace' means adopting a SaaS solution. - The tools: Azure Migrate (assessment), Azure Site Recovery (replication), Azure Database Migration Service (database migration).
Edge Cases and Tricky Distinctions - If a question says 'migrate a SQL Server database to Azure,' is it refactor or rehost? If you move the database to a SQL Server VM, it's rehost. If you move to Azure SQL Database, it's refactor. - For web apps: moving from IIS on a VM to Azure App Service is refactor, not rehost. - If the application is containerized and moved to AKS without code changes, it's still considered rehost (or sometimes 'recontainerize' but that's not in the 5 Rs).
Memory Trick Use the acronym '5 R's' and associate each with a level of change:
Rehost = change nothing (0)
Refactor = change platform (1)
Rearchitect = change architecture (2)
Rebuild = change everything (3)
Replace = change to SaaS (outsource)
Another trick: Think of the effort scale: Rehost (lowest effort) -> Refactor -> Rearchitect -> Rebuild -> Replace (highest effort, but not always; Replace can be low effort if SaaS is ready).
The five migration strategies are Rehost, Refactor, Rearchitect, Rebuild, and Replace (5 Rs).
Rehost (lift-and-shift) moves applications unchanged to Azure VMs; it's the fastest but least optimized.
Refactor (replatform) makes small changes to leverage PaaS services like Azure App Service or Azure SQL Database.
Rearchitect modifies the application architecture (e.g., to microservices) to improve scalability and resilience.
Rebuild rewrites the application from scratch using cloud-native technologies like Azure Functions and Cosmos DB.
Replace adopts a SaaS solution (e.g., Exchange Online) instead of migrating the existing application.
Azure Migrate is the primary tool for discovery, assessment, and cost estimation during migration planning.
Azure Site Recovery is used for VM replication and failover in rehost migrations.
Azure Database Migration Service supports database migrations with minimal downtime.
The choice of strategy affects total cost of ownership (TCO), with rehost having lower initial cost but higher operational cost.
Each workload should be evaluated independently; a mix of strategies is common in real-world migrations.
Post-migration optimization (right-sizing, auto-scaling, cost management) is critical to realize cloud benefits.
These come up on the exam all the time. Here's how to tell them apart.
Rehost (Lift-and-Shift)
No code changes required
Moves VMs as-is to Azure IaaS
Fastest migration time
No cloud-native benefits (auto-scaling, managed services)
Higher long-term operational overhead
Refactor (Replatform)
Minor code changes may be needed
Moves to PaaS services (App Service, Azure SQL DB)
Moderate migration time
Gains cloud benefits like auto-scaling and managed backups
Lower long-term operational overhead
Mistake
Rehost is always the cheapest migration strategy.
Correct
Rehost often has the lowest initial migration cost but can lead to higher ongoing operational costs because you still manage OS, patching, and scaling. Refactor or rearchitect may reduce long-term costs by using PaaS services that include management.
Mistake
Rearchitect and Rebuild are the same thing.
Correct
Rearchitect modifies the existing application architecture (e.g., breaking a monolith into microservices) but reuses much of the existing code. Rebuild involves rewriting the application from scratch, often using a different technology stack.
Mistake
You must use Azure Migrate for all migrations.
Correct
Azure Migrate is the recommended tool for discovery and assessment, but you can migrate manually or use other tools like Azure Site Recovery, Azure Database Migration Service, or even third-party tools. The exam expects you to know Azure Migrate as the primary assessment tool.
Mistake
The 'Replace' strategy means you replace the application with a custom-built solution.
Correct
Replace means adopting a SaaS solution (e.g., Dynamics 365, Office 365) instead of running your own application. It does not involve building a new application.
Mistake
All workloads must follow the same migration strategy.
Correct
Each workload should be evaluated individually. A company may use rehost for some applications, refactor for others, and replace for a few. The exam often tests this by describing a scenario with multiple workloads.
Rehost (lift-and-shift) moves applications to Azure without any changes, typically by migrating VMs as-is. Refactor (replatform) involves making small modifications to take advantage of PaaS services, such as moving a web app from IIS on a VM to Azure App Service. The key distinction is that refactor requires some development effort to change the platform layer, while rehost does not. On the exam, if the question says 'minimize changes' or 'fast migration,' choose rehost. If it says 'reduce management overhead' or 'use managed services,' choose refactor.
Choose rearchitect when you want to improve scalability, resilience, or deployment speed but still want to preserve existing business logic and code. Rearchitect modifies the architecture (e.g., breaking a monolith into microservices) but reuses most of the codebase. Choose rebuild when the existing application is outdated, unsupported, or cannot meet new requirements, and you are willing to rewrite from scratch. Rebuild gives you the most cloud-native benefits but at the highest cost and risk. On the exam, look for phrases like 'modify the architecture' vs 'rewrite the application.'
The Replace strategy means you stop using your existing application and instead adopt a Software as a Service (SaaS) solution that provides similar functionality. For example, replacing an on-premises Exchange Server with Exchange Online (Office 365) or replacing a custom CRM with Dynamics 365. This strategy eliminates the need to migrate the application itself, but it may require data migration and user training. On the exam, Replace is often the correct answer when the scenario mentions 'use a cloud-based service' or 'SaaS alternative.'
Azure Migrate is the primary tool for discovering on-premises servers, applications, and dependencies. It provides a centralized hub for assessment and migration planning. You install a lightweight appliance that collects performance data and software inventory. Azure Migrate also integrates with Azure Site Recovery and Azure Database Migration Service for actual migration. On the exam, if a question asks about 'assessing on-premises workloads for migration,' the answer is Azure Migrate.
Yes, absolutely. In fact, most real-world migrations use a combination of strategies. For example, a company might rehost legacy applications with no source code, refactor modern web apps to use App Service, and replace email with Office 365. The exam often presents a scenario with multiple workloads and asks you to select the appropriate strategy for each. Remember that each workload should be evaluated independently based on business and technical requirements.
Azure Site Recovery (ASR) is used primarily for rehost migrations. It replicates on-premises VMs to Azure, allowing you to failover and bring them online in Azure with minimal downtime. ASR supports both physical and virtual servers and provides continuous replication. It is also used for disaster recovery. On the exam, ASR is associated with 'rehost' or 'lift-and-shift' migrations. For database migrations, use Azure Database Migration Service instead.
The migration strategy directly impacts both short-term and long-term costs. Rehost has low initial migration cost but high ongoing operational costs (OS licenses, patching, management). Refactor reduces operational costs by using PaaS services that include automatic patching, backups, and scaling. Rearchitect and rebuild have higher upfront development costs but can significantly lower operational costs and improve efficiency. Replace converts capital expenditure (hardware, licenses) into operational expenditure (subscription fees). On the exam, you may be asked to identify which strategy leads to the lowest TCO over time, typically refactor or rebuild.
You've just covered Cloud Migration Strategies Overview — now see how well it sticks with free AZ-900 practice questions. Full explanations included, no account needed.
Done with this chapter?