The 7 Rs are workload-level choices for deciding what to do with an application or service during a cloud migration: Retire, Retain, Rehost, Relocate, Repurchase, Replatform, and Refactor (or rearchitect). They are planning categories, not a mandatory sequence. Different workloads in the same portfolio can—and often should—take different paths.
A choice should reflect the workload’s business value, dependencies, constraints, urgency, and the value the organization expects after the move. AWS describes a migration strategy as “the approach used to migrate a workload into the AWS Cloud.” AWS Prescriptive Guidance and Microsoft’s Azure Cloud Adoption Framework use related but different taxonomies, so identify the provider framework being applied.
The 7 Rs at a glance
| Strategy | What it means | When it can fit | Main caution |
|---|---|---|---|
| Retire | Decommission or archive the workload. | It is redundant, obsolete, unused, or no longer worth operating. | Check owners, interfaces, data-retention duties, scheduled jobs, and downstream dependencies first. |
| Retain | Keep it in its current environment for now. | Compliance, residency, specialized hardware, dependencies, timing, or risk makes migration unsuitable. | Document the reason and a review trigger so an exception does not become permanent by accident. |
| Rehost | Move it largely unchanged (“lift and shift”). | It is stable and compatible, and speed or minimal disruption matters most. | Existing performance, reliability, architecture, cost, and technical-debt problems move with it. |
| Relocate | Move servers or infrastructure to an equivalent cloud platform without rewriting the application. | A platform-level move preserves the operating model while changing where it runs. | Validate account, region, network, governance, and target-platform constraints. |
| Repurchase | Replace the current application or license with another product, often SaaS. | An adequate cloud service can replace a product that is costly or unsuitable to maintain. | Compare features, data handling, security, compliance, integrations, licensing, and exit options. |
| Replatform | Move it with bounded hosting or platform improvements (“lift, tinker, and shift”). | Managed services, an operating-system upgrade, or containers can improve operations without a major rewrite. | Set a firm change boundary; major architectural work is refactoring, not ordinary replatforming. |
| Refactor/rearchitect | Change code or architecture to use cloud-native capabilities. | The expected gains in agility, scale, performance, or resilience justify substantial effort. | It has the greatest delivery complexity; avoid combining it with a time-critical migration without a clear business case. |
AWS provides the detailed definitions in its cloud migration strategy overview and Prescriptive Guidance.
What each R means in practice
1. Retire
Retirement is a valid migration outcome when a workload no longer provides enough business value. Do not equate low resource use with permission to shut it down. Confirm the business owner, integrations, reports, audit obligations, retention rules, and recovery requirements.
The Tool Desk
Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →#1 Best Overall
AWS Prescriptive Guidance gives screening examples: average CPU and memory below 5% for 90 days may indicate a “zombie” application; 5%–20% over 90 days may indicate an “idle” application; and no inbound connection for 90 days may be a retirement signal. These are AWS examples, not universal thresholds or automatic shutdown rules. Check seasonal operation and whether the 90-day period represents a normal business cycle.
2. Retain
Retain means deliberately leaving a workload where it is, at least temporarily. Reasons can include data-residency or compliance requirements, an unavailable cloud equivalent, specialized hardware, difficult dependencies, high migration risk, or an unfavorable schedule. Record who owns the exception, what would change the decision, and when it will be reviewed.
Rank #2
3. Rehost
Rehosting transfers a workload with little or no application change. It is often suitable for a stable system when the immediate goal is datacenter exit, faster delivery, or lower disruption. Because the code and architecture remain substantially the same, rehosting alone does not repair an application’s defects or guarantee lower cost, better resilience, or improved performance.
4. Relocate
Relocation is a platform-level move: a group of servers or infrastructure moves to a cloud version of an equivalent environment while the application remains substantially intact. This can preserve familiar tools and operating procedures. Validate target-platform support, network design, identity, governance, account structure, region availability, and licensing before committing.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Rank #3
5. Repurchase
Repurchase replaces the existing product or licensing model, commonly with a SaaS offering. Evaluate the replacement as a business system, not just a hosting option. Confirm required features, data export and deletion, security controls, regulatory obligations, identity integration, APIs, service-level commitments, total licensing cost, and a practical exit plan.
6. Replatform
Replatforming keeps the application’s essential design but makes limited changes to improve its operating environment. Examples include moving SQL Server to Amazon RDS for SQL Server, adopting a managed database, upgrading the operating system, or packaging an application in containers. Define what is in scope; if the work changes core boundaries, data flows, or business logic, classify it as refactoring or rearchitecting.
7. Refactor or rearchitect
Refactoring changes the code or architecture to obtain cloud-native benefits such as elastic scale, faster releases, or greater resilience. It can be the right answer for a high-value product whose current design blocks an important business objective. It also demands the most engineering time, testing, organizational capacity, and delivery control. AWS notes that large migration programs may move workloads first and modernize them afterward rather than making every migration a major rewrite.
How to choose the right strategy for an application
- Define the business driver. State whether the goal is datacenter closure, operating-cost reduction, resilience, faster delivery, scale, risk reduction, or a new capability. Microsoft recommends connecting the strategy to the gap between the current and desired state.
- Build a trustworthy inventory. Capture the application owner, users, infrastructure, performance, data, lifecycle, support model, and business criticality. AWS’s detailed portfolio discovery guidance treats this information as the basis for assigning strategies.
- Map dependencies and constraints. Record upstream and downstream systems, integrations, data-residency rules, specialized hardware, security obligations, operational responsibilities, team skills, and technical complexity.
- Compare effort, risk, and intended value. Rehost generally minimizes application change; replatform adds bounded platform improvement; refactoring adds substantial modernization work; repurchasing changes the product and licensing model; retiring or retaining may be the most rational outcomes.
- Assign a migration wave. Group workloads according to interdependencies and complexity rather than treating each application as isolated. A strategy that works for one service may be unsafe for a tightly coupled neighbor.
- Revisit the decision. Portfolio plans mature as discovery improves and migration waves produce evidence. Reassess when ownership, usage, compliance requirements, target services, or business priorities change.
Decision criteria when several Rs seem possible
Use the same criteria for each candidate and document the trade-off:
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Scan for outdated or missing drivers - takes under a minuteDriver Scan →Best Value
- Business value and urgency: how important is the workload and how quickly must the outcome arrive?
- Application change: can the system move unchanged, tolerate bounded platform changes, or require a redesign?
- Delivery effort and risk: what skills, testing, downtime planning, and rollback capability are available?
- Future operating model: who will run it, and how will hosting, support, licensing, and observability work afterward?
- Security and compliance: are residency, control, audit, or data-handling requirements satisfied?
- Dependencies and integrations: can connected systems move together, or must this workload wait?
- Time to migrate versus time to modernize: would combining both goals create unacceptable schedule or delivery risk?
These criteria synthesize provider guidance; AWS and Microsoft do not prescribe a single universal scoring formula.
A simple portfolio example
Imagine three workloads. A redundant reporting application may be a retire candidate after owners, dependencies, scheduled jobs, and retention obligations are checked. A stable internal service with no near-term modernization need could be rehosted for a fast move or retained if a constraint blocks migration. A high-value product whose architecture prevents the required scale may justify refactoring when the expected business value supports the cost and delivery capacity. These are applications of the provider definitions, not a prescription for every estate.
AWS and Azure do not use identical 7-R lists
The list in this article follows AWS’s named seven: Retire, Retain, Rehost, Relocate, Repurchase, Replatform, and Refactor/rearchitect. Microsoft’s Azure framework uses a broader, differently divided set: Retire, Retain, Rehost, Replatform, Refactor, Rearchitect, Rebuild, and Replace. “Repurchase” and “Replace,” or “Refactor” and “Rearchitect,” may therefore refer to different boundaries depending on the provider. Name the framework in portfolio documents so teams do not compare unlike categories.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
Free tools Windows power users keep installed
One-click scans. No signup required.
