Why do people still think databases should not run on Kubernetes?

What people? Have you read or listened to people “still thinking” this who don’t give a reason? Do you have some reasons people should not still think this?

I have never needed containers or Kubernetes, but as a long-time DBA and programmer I can speculate. Relational database engines generally perform best running close to the hardware, because of optimizations around memory and I/O. Every layer of abstraction introduced adds drag that affects all clients of the database.

Edit: This on HN just now. Apparently some people don’t still think databases shouldn’t run on K8S.

https://kubeblocks.io/