Comment by Sébastien Portebois on "Any service which lets you replicate your services over AWS, GCP and Azure + replication over multiple regions in each provider?"

I never looked for a managed solution for this, but my approach if I had to prepare to manage this to reduce the locked-in would be:

use Kubernetes to manage the workloads, then the deploys#update are as easy as platform-agnostic kubectl or helm calls
either use Terraform or similar product to manage the underlying infrastructure (which let you manage aws, azure, gce, ...)
or (ideal case but more work) consider Mesosphere / DC/OS to even add one more level of abstraction under kubernetes.

There are real big challenges there, and we're only considering the payloads here, but you should also consider if your code directly uses S3 (which is common), or similar other app-level dependencies to AWS tech.

Finally another real challenge is to know and mitigate your third-party dependency to AWS if that's really what you need to achieve.

If the goal is resiliency rather than avoiding locked-in, I'd recommend considering multi-region HA rather than multi-cloud provider.

Search Hashnode