Cloud Migration Checklist: Moving to AWS, Azure or Google Cloud Without Downtime
A practical sequence for moving applications to the cloud: assess, choose a strategy, build the landing zone, migrate in waves, cut over safely and control the bill.

Cloud migrations rarely fail because of the cloud. They fail because of what nobody wrote down: a scheduled job on an old server, a hard-coded IP address, a database that is larger than anyone thought, or a cutover planned for a Friday afternoon. This checklist is the sequence we follow to avoid those surprises.
1. Assess what you have
- Inventory every application, database, scheduled job and integration, including the ones only one person knows about.
- Map dependencies: which services call which, and what breaks if one moves first.
- Measure real usage: CPU, memory, storage, traffic and peak times, so you size the new environment from data rather than guesses.
- Note constraints: data residency, compliance requirements and maintenance windows.
2. Choose AWS, Azure or Google Cloud
All three are mature. The right choice usually follows your existing skills and contracts more than feature lists:
| Provider | Often the best fit when… |
|---|---|
| AWS | You want the widest range of services and the largest hiring pool |
| Azure | You already rely on Microsoft 365, Active Directory or .NET |
| Google Cloud | Data analytics, BigQuery and Kubernetes are central to your plans |
Check that your required regions exist in your customers' markets, for example in the UAE, India, Canada, the UK or specific US regions, if data residency matters.
3. Pick a strategy per application
- Rehost (lift and shift): move as-is to virtual machines. Fastest, least change, least benefit.
- Replatform: small changes such as moving to a managed database or containers. Often the best balance.
- Refactor: redesign for cloud-native services. Most benefit, most effort; save it for systems that justify it.
- Retire or replace: some applications should be switched off or replaced with a SaaS product instead of moved.
Deciding per application keeps the project moving: not everything deserves a rewrite.
4. Build the landing zone first
Before moving anything, set up the foundations:
- Separate accounts or projects for production, staging and shared services.
- Networking, VPN or private connectivity to existing systems.
- Identity and access with single sign-on and least privilege.
- Logging, monitoring and alerting.
- Budgets and cost alerts.
- Everything defined as code with Terraform or a similar tool, so environments are repeatable.
5. Containerise and automate
Containers with Docker make applications portable and environments consistent. Add a CI/CD pipeline that builds, tests and deploys every change automatically. This is the point where a migration also becomes a DevOps improvement: fewer manual steps, fewer deployment mistakes and faster releases. Kubernetes or a managed container service is worth it when you run many services; for a few, simpler managed platforms are often enough.
6. Migrate data safely
- Choose between a one-time copy during a maintenance window and continuous replication for large or busy databases.
- Test the migration on a copy first and measure how long it takes.
- Validate row counts and checksums after every run.
- Keep the old database read-only and available until the new one is proven.
7. Cut over without downtime
- Lower DNS time-to-live values a few days before cutover.
- Run the new environment in parallel and test it with real traffic patterns.
- Switch traffic gradually where possible, or in a planned window with a clear rollback plan.
- Watch errors, latency and business metrics closely for the first days.
- Decommission old servers only after a stable period.
Plan cutovers early in the week with the full team available, never just before a weekend or holiday.
8. Security and compliance
- Encrypt data at rest and in transit and manage secrets in a vault, not in code.
- Restrict network access and close unused ports.
- Enable audit logging and keep it for the period your regulations require.
- Apply the data-protection terms for your markets, for example UK GDPR, PIPEDA, the UAE PDPL or US state privacy laws.
9. Control the bill
Cloud costs grow quietly. From the first week:
- Right-size instances based on measured use.
- Turn off non-production environments outside working hours.
- Use savings plans or committed-use discounts for steady workloads.
- Tag resources by team and environment so costs are visible.
- Review spend monthly and remove idle resources.
Our cloud cost audit article shows where savings usually hide.
Timeline and cost
A typical cloud and DevOps engagement at Blurbee runs three to ten weeks, at $26 to $52 per hour depending on seniority. Local agency rates for comparison, as published on our location pages:
| Market | Typical local agency rate (per hour) |
|---|---|
| United States | $75–$220 |
| United Kingdom | £45–£130 |
| United Arab Emirates | AED 120–420 |
| Canada | CA$60–175 |
| India | ₹800–3,600 |
Testing before cutover
Testing is where migration risk is retired. Before switching any real traffic:
- Functional tests of every critical journey, ideally automated so they can be repeated.
- Performance tests at expected and peak load, comparing response times with the old environment.
- Failure tests: stop a server or service on purpose and confirm the system recovers.
- Backup and restore tests: restore a backup into a clean environment and check that it works.
- Security checks: open ports, access rights and exposed secrets.
Write down what "good" looks like for each test before you run it, so the go or no-go decision is objective.
Who does what during the migration
Clear roles prevent confusion on cutover day. Typically you need an owner who makes the go or no-go call, engineers for each application and database, someone watching monitoring and business metrics, a person handling communication with users and stakeholders, and a rollback owner who knows exactly how to switch back. Agree these roles and a communication channel before the day.
After the migration
The work does not end at cutover:
- Keep monitoring closely for at least two weeks.
- Right-size resources once real usage data is available.
- Remove temporary migration tools and old access.
- Update documentation, runbooks and architecture diagrams.
- Decommission the old environment only after a stable period and a final backup.
- Hold a short review of what went well and what to change next time.
Common migration mistakes
- Migrating everything at once instead of in waves.
- Forgetting scheduled jobs, email relays or hard-coded IP addresses.
- Underestimating data transfer time for large databases.
- Leaving costs unmanaged until the first large bill arrives.
Costs to plan for during the migration
Running two environments in parallel for a period is normal and should be in the budget. Data transfer out of the old environment, temporary replication tools, extra testing environments and additional support during cutover all add cost for a few weeks. Planning for them avoids pressure to rush cutover just to stop paying twice, which is when mistakes happen.
A one-page cutover checklist
- Rollback plan written, tested and owned by a named person.
- DNS time-to-live lowered in advance.
- Final data sync scheduled and its duration measured.
- Monitoring dashboards and alerts ready for the new environment.
- Support team briefed and users informed of any maintenance window.
- Go or no-go criteria agreed before the day.
Key takeaways
Inventory everything before moving anything, choose a strategy per application rather than one for the whole estate, build the landing zone as code, migrate in waves, test restores and failures before cutover, and control costs from the first week. A calm, well-planned cutover early in the week beats a rushed one every time.
Getting started
Begin with an infrastructure assessment: an inventory, a dependency map and a prioritised plan. That alone often reveals quick wins. See our cloud and DevOps services or request an assessment.