Can you use Digital Ocean managed Kubernetes to deploy on GCP or AWS? Or is there a provider tie-in?

If the answer is no, you can’t, then:

How easy is it to move the running cluster to a different zone or a cloud provider?

Are there any pitfalls here in this regard?

If the answer is no, then

Basically, it’s all going to be manual. You are pretty much going to set up everything again from scratch somewhere else. Since Federation V2 is not yet ready for production use, multicluster communication is always going to be an issue.

Can you easily replicate the cluster somewhere else by exporting the configuration used by the managed Digital Ocean Kubernetes master, and deploy it on, for example, AWS? If this is universal then I think by prebuilding and bootstrapping components as custom charts and enable database concurrency and replicating it before migrating the cluster over and pull the plug for the cluster here would reasonably work?

What is the best practice for delegating the control of the master node configuration?

Subconsciously, I have a hunch that having a managed solution is a bad idea because if you don’t control the initial deployment, then it will not be replicable without the original deployment manager, which is not portable, not standardized and often proprietary, which defeats the purpose of Kubernetes, which is to standardize deployment and enable vendor neutrality.

So if I am not missing anything, then you should always configure your own master node using only first-party tools?

That is deploying everything yourself using kubectl on vanilla droplets? That can maximize portability even without federation V2 support. The Kubenete idea is to use infrastructure as a service, or IaaS. Just like Mobile service carriers or ISPs, they should be a pipe for data aka. IaaS, not a platform as a service PaaS. So, moving from T&T to T-Mobile, for example, should be frictionless.

So, is Digital Ocean managed Kubernetes a form of PaaS?