We're doing pretty well on Digital Ocean so far, but it's not wise to let ourselves be coupled to the provider without some contingency plan. Let's test out how we might apply our gitlab-cloud-runner terraform config to other providers such as:
- GKE
- AWS EKS
- WMF OpenStack?
Let's implement submodules for one or two of the above in our gitlab-cloud-runner terraform and see if we can get it to cleanly provision. All provider submodules should share the same interface, that is they should reference the exact same set of input variables and provide the same outputs.