AWS/Azure to GDC: what the migration involves.
Google Distributed Cloud, including its air-gapped configurations, is pulling government workloads. Service mapping, compliance re-inheritance, and the playbook that keeps a migration from becoming a rewrite.
Google Distributed Cloud (GDC) has become a serious destination for government workloads, including its air-gapped configurations for classified environments. That has teams running on AWS GovCloud or Azure Government asking a practical question: what does it take to move? It is more work than a repoint and less than a datacenter migration, and how you built decides where in that range you land.
What GDC is, in one paragraph
GDC is Google's portfolio for running Google Cloud infrastructure outside Google's public regions: connected configurations for sovereign and edge requirements, and air-gapped configurations for classified missions with no external connectivity. It is Kubernetes-native throughout: the platform primitives are GKE-based, which is why containerized workloads migrate cleanly and VM-shaped legacy workloads don't.
The migration, service by service
| You run today | You land on | Friction |
|---|---|---|
| EKS / AKS clusters | GKE / GDC Kubernetes | Low: manifests mostly port |
| EC2 / Azure VMs | GDC VM runtime or containerize first | Medium-high: refactor pays off |
| S3 / Blob Storage | Object storage (S3-compatible APIs) | Low: mind egress and data gravity |
| IAM policies | Google IAM model | Medium: different philosophy, full remap |
| CloudFormation / ARM | Terraform on Google providers | Low if you're already Terraform |
| Managed databases | Platform equivalents or self-managed | Medium: inventory features before committing |
Compliance re-inheritance
Your authorization doesn't travel with your workload. Control inheritance maps rebuild against the new platform's posture, evidence collection re-points at new APIs, and the boundary diagram gets redrawn. This is where accelerated-ATO practices pay off: if your evidence is generated by pipelines rather than typed into documents, re-homing it is a configuration job.
The playbook that works
Containerize first, before the move, on your current cloud; it de-risks everything after. Terraform everything so environments are declarative and portable. Move stateless services first, data second, stateful legacy last. Re-map identity early: IAM is where migrations stall. Run the authorization work as a parallel track starting on day one.
Quick answers
Do we lose our ATO when we migrate to GDC?
Is GDC only for air-gapped classified work?
How long does an AWS/Azure to GDC migration take?
Related reading
Working on this now?
Tell us where your program stands and we'll tell you what we would do next.
Talk to Ausper