Managed Database Pricing: RDS vs Azure SQL vs Cloud SQL

How Managed Database Pricing Works

AWS RDS, Azure SQL Database, and Google Cloud SQL all charge along the same three axes: compute (the instance size running the database engine), storage (provisioned disk plus IOPS), and data transfer. The complexity is that each provider bundles these differently — RDS and Cloud SQL price compute and storage as separate line items, while Azure SQL Database's DTU and vCore tiers bundle a fixed amount of storage and IOPS into the compute price, making direct comparison harder than it first appears.

All three providers also offer a Bring Your Own License option alongside License-Included pricing for SQL Server and Oracle engines. License-Included pricing folds Microsoft or Oracle licensing fees into the hourly instance rate and is usually 1.5–3x more expensive than the equivalent open-source engine (MySQL, PostgreSQL) on the same instance size — the licensing cost, not the infrastructure, is what drives commercial database TCO.

Instance-Based Pricing Compared

For a comparable mid-size production database (4 vCPU, 16GB RAM, general-purpose SSD storage, single-AZ, PostgreSQL/MySQL-equivalent engine), approximate on-demand monthly costs look like this:

Compute only, US region, single-AZ, excludes storage, IOPS, backup and licensing. Reserved/committed pricing reduces all three by 35–55% for 1–3 year terms.

Serverless and Auto-Scaling Pricing

All three providers now offer serverless database tiers that bill per second of actual compute consumed rather than a fixed instance size: Aurora Serverless v2 (AWS), Azure SQL Database serverless, and Cloud SQL's edition-based auto-scaling. These are billed in Aurora Capacity Units (ACUs) or vCore-seconds and can cut costs 40–60% for workloads with intermittent or spiky traffic — dev/test databases, internal tools, and low-traffic APIs — but are usually more expensive than a right-sized provisioned instance for databases under continuous, predictable load.

Storage, IOPS and Backup Costs

Storage is billed separately from compute on AWS and GCP: general-purpose SSD storage runs $0.08–0.115/GB-month, with provisioned IOPS storage costing extra per IOPS above the baseline included with general-purpose volumes. Automated backups are free up to a size equal to your total provisioned database storage; beyond that, backup storage bills at the same per-GB rate as primary storage — a common surprise for teams with long retention windows or frequent manual snapshots.

High Availability and Read Replica Costs

Multi-AZ deployments (AWS RDS Multi-AZ, Azure SQL Business Critical/zone-redundant, GCP Cloud SQL HA) roughly double the compute cost of a single-AZ instance because a full standby replica runs continuously in a second availability zone. Read replicas for scaling read traffic are billed as additional full-price instances — three read replicas of a $490/month primary add another $1,470/month regardless of how lightly they're queried, since you pay for provisioned capacity, not usage.

License-Included vs Bring Your Own License

For SQL Server and Oracle workloads, License-Included pricing on RDS or Azure can cost 2–3x the equivalent Linux/open-source instance, since it bundles per-core Microsoft or Oracle licensing fees. Organizations with existing Microsoft Enterprise Agreements or Oracle licenses can use BYOL pricing instead, paying only for infrastructure — but BYOL requires the organization to separately track license compliance (core counts, Software Assurance) rather than having the cloud provider manage it, which adds administrative overhead that must be weighed against the savings.