A managed database service — RDS, Cloud SQL, Azure Database — costs more per compute-hour than a self-managed database on an equivalent VM. The unit price is higher; the operational overhead is lower. Whether the total cost of ownership is lower depends on the workload and the engineering organisation.
For small teams with limited database administration capability, managed services consistently win on total cost even at the higher unit price. Patches, backups, replication, failover — all of these require skilled engineering time when self-managed, and the cost of that time at market rates exceeds the platform premium at most workload sizes relevant to mid-market organisations.
For large teams with strong DBA capability and very high-scale workloads, the economics can flip. At sufficiently high scale, the control available through self-management can produce better cost-per-query outcomes than the managed equivalent. This is a genuinely rare case for organisations below the very top of the enterprise scale.
The more interesting economic argument for platform services is operational resilience rather than cost. A managed database service includes multi-AZ replication, automated failover, point-in-time recovery, and automated minor-version patching as defaults. Replicating these capabilities with self-managed infrastructure requires significant engineering investment.
Our FinOps reviews regularly identify IaaS workloads that would be cheaper overall on managed services when engineering overhead is included in the comparison, and managed service workloads that would be cheaper on IaaS when scale justifies the self-management investment.
The practical upshot: if you are making an IaaS-versus-PaaS decision, model the engineering overhead explicitly and include it in the cost comparison. The list-price comparison is not the right comparison.