I have a client who's pretty massive database is pretty damn critical, not only for revenue but also legal reasons. It is managed by a separate team. All applications however run on k8s (well, actually openshift), the main reason being organizational scaling of development teams. With the current setup in place, increasing the amount of dev teams we'd need to support would be pretty painless. We do however run certain smaller (<200gb) databases in-cluster where performance is less of a focus, which, because of the storage abstractions currently in place could otherwise become a potential bottleneck.
