Hey Kubernetes practitioners, one of the questions that keeps coming up is how large teams are using Kubernetes? Do you use namespace as a ownership boundary in your teams? Do you share Kubernetes cluster among your team by configuring hierarchical ownership policies? Can you share your experience?

Why this question?

The Litmus team is considering chaos at namespace level. You will be able to run the complete chaos infrastructure within the namespace.

Here is the description of a possible scenario:

Kubernetes cluster is setup on one of the cloud service providers like EKS, GKE, AKS or DOKS where the management of the cluster is not managed by the team. Then your team has a set of SREs or admins who have cluster wide access through service accounts to help manage the administration of applications and the cluster itself. When a developer wants the Kubernetes environment, a new namespace is created with service account settings with access to that developer. The developer has enough levers within the namespace and gets the Kubernetes environment that is required for development needs.

Are there any scenarios in which you felt there are limitations? Is this a common practice? Or developers are better off with their own small clusters?