How we help / Cloud & DevOps / Cloud migration
Cloud migration services for AWS, Google Cloud and Azure
We move applications, data and databases between clouds, or from your own servers into one, and keep the business running while we do it.
Most of our migration work is one of three moves: Google Cloud to AWS, AWS to Google Cloud, or on-premises to the cloud. We plan the move, do the engineering and stay through cutover, so your team is not left with a plan and no one to run it.
Migrations we take on
Each move has its own service mapping and data path. These are the ones people bring us most.
- Google Cloud to AWS
- Map GCP services to their AWS equivalents, move Cloud Storage buckets to S3, and rebuild identity, networking and pipelines on AWS.
- AWS to Google Cloud
- The same move in the other direction: S3 to Cloud Storage with Storage Transfer Service, and workloads onto GKE, Cloud Run or Compute Engine.
- On-premises to cloud
- Move applications off your own servers, rehosted as they are or containerized on the way when they need to scale.
- Data and storage migration
- Copy object storage and file shares at scale with verification and incremental syncs, so the final copy at cutover is small.
- Database migration
- Replicate databases with change data capture so the old and new copies stay in step until you switch.
- Multi-cloud and hybrid
- Run some workloads in more than one cloud, or keep part of the estate on premises, with one way to deploy and monitor it all.
How a migration runs
Six steps, in this order. Each wave of workloads goes through the last four, so problems show up on a small move first.
- Assess what you run
Inventory servers, services, data and the dependencies between them, with the provider’s own discovery tools where they help, and estimate cost on the target cloud.
- Plan the target and the waves
Choose rehost, replatform or refactor for each workload, design the landing zone in code, and group workloads into waves that can move together.
- Move one workload first
Migrate a small, real workload end to end to prove the network, identity, pipelines and monitoring before the larger waves.
- Copy and sync the data
Run the bulk copy early, keep it current with incremental syncs or replication, and verify object counts and checksums.
- Cut over with a way back
Freeze writes, run the final sync, switch DNS and traffic, and keep the old environment ready until the new one has run cleanly.
- Tidy up and right-size
Remove the old resources, size instances to real load, and hand over runbooks and dashboards to your team.
Tools we use
Mostly the providers’ own migration services, plus infrastructure as code so the new environment can be rebuilt.
- Assessment
- Google Cloud Migration Center, AWS Application Discovery Service and AWS Transform, Azure Migrate.
- Storage
- AWS DataSync, Google Storage Transfer Service, rclone for smaller buckets and file shares.
- Databases
- AWS Database Migration Service, Google Database Migration Service, native replication.
- Infrastructure and delivery
- Terraform, Docker, Kubernetes, Helm, GitHub Actions, GitLab CI and ArgoCD.
Questions people ask first
How long does a cloud migration take?
It depends on how much you run and how tangled it is. A single application with its data usually moves in weeks; a whole estate is planned in waves over several months. The assessment gives you a schedule built from your own inventory, not a guess.
How much downtime should we expect?
For most workloads, only the cutover window. Data is copied and kept in sync beforehand, so at cutover there is just a final sync, a check and a traffic switch. We agree the window and the rollback plan with you before the day.
What does it cost to move data out of Google Cloud or AWS?
Data leaving a cloud is normally charged as egress. Google Cloud, AWS and Azure each offer to waive those fees for customers moving data off their platform, on request and with conditions, so we check eligibility before the copy is planned. For very large datasets we also compare network transfer with offline transfer devices.
Should we lift and shift or re-architect?
Usually both, workload by workload. Rehosting is fastest and lowest risk; re-architecting pays off where an application has to scale or fails often. We decide per workload during planning and say why.
Do you only plan the migration, or carry it out too?
Either. Some teams want an assessment and plan they run themselves; most ask us to do the engineering and stay through cutover. After the move we can hand over completely or keep running and improving the new environment with you.
Where we have done it
Guides from our blog
Planning a move?
Tell us what you run today and where it needs to go. We will reply with what we would look at first.