Do we mean managed or PG on K8s like CNPG? In all cases, I use the infra to simplify things like having disk redundancy and failover nodes, not because 12GB is interesting.