GCP AlloyDB cost: compute per vCPU-hour plus regional storage

Quick answer

AlloyDB bills compute per vCPU-hour and memory for its primary and read-pool instances, plus regional storage per GB-month (with no separate per-I/O charge, unlike Aurora-style engines) and backups. It is PostgreSQL-compatible and aimed at demanding transactional and analytical workloads. For smaller or simpler workloads, Cloud SQL is cheaper; AlloyDB earns its cost at scale and performance.

AlloyDB is Google Cloud's PostgreSQL-compatible database built for demanding workloads. Its pricing separates compute from storage, and notably does not charge per I/O the way some Aurora-style engines do, which changes the calculus for read-heavy workloads.

Compute plus storage

Dimension	Bills for
Compute	Per vCPU-hour and memory, for primary and read-pool instances
Storage	Regional, per GB-month, no separate I/O charge
Backups	Per GB-month of backup storage

You pay for the compute of the primary instance and any read-pool instances by vCPU-hour and memory, and for regional storage per gigabyte-month. Because there is no separate per-I/O charge, a read-heavy workload does not pay per query the way it would on a per-I/O engine, which can make AlloyDB more predictable for such workloads.

Read pools scale reads

AlloyDB separates the primary (writes) from read-pool instances (reads), so you add read capacity by adding read-pool nodes, each billed for its compute. This is the AlloyDB version of the read-replica tradeoff: read pools cost compute but offload the primary and scale read throughput.

AlloyDB vs Cloud SQL

For smaller, simpler PostgreSQL workloads, Cloud SQL is cheaper and sufficient. AlloyDB earns its higher cost on demanding transactional and analytical workloads that need its performance, scale, and read pools. Choosing AlloyDB for a modest database is overspend; choosing Cloud SQL for a workload that needs AlloyDB is a performance ceiling.

Controlling AlloyDB cost

Right-size the primary instance to real load, add read-pool nodes only as read throughput requires, manage backup retention, and use the standard workload levers, indexing, caching, and query tuning, so the instance you need is smaller. The absence of a per-I/O charge means the lever is compute sizing, not query count, so right-sizing the instances is the main control.

FAQ

How is GCP AlloyDB priced?

By compute per vCPU-hour and memory for its primary and read-pool instances, plus regional storage per GB-month with no separate per-I/O charge, plus backup storage. The absence of a per-I/O charge distinguishes it from Aurora-style engines and can make read-heavy workloads more predictable.

Is AlloyDB cheaper than Cloud SQL?

For smaller or simpler PostgreSQL workloads, no, Cloud SQL is cheaper and sufficient. AlloyDB costs more but earns it on demanding transactional and analytical workloads that need its performance, scale, and read pools. Match the engine to the workload's demands.

Does AlloyDB charge per I/O?

No. Unlike some Aurora-style engines, AlloyDB does not have a separate per-I/O charge; you pay for compute and regional storage. This means a read-heavy workload does not accrue per-query I/O cost, so compute sizing rather than query count is the main cost lever.

How do AlloyDB read pools affect cost?

Read-pool instances add read capacity by running additional nodes, each billed for its compute. They offload reads from the primary and scale read throughput, the same tradeoff as read replicas: they cost compute but improve read performance. Add them only as read throughput requires.

How do I reduce AlloyDB cost?

Right-size the primary instance to real load, add read-pool nodes only as read throughput requires, manage backup retention, and reduce workload with indexing, caching, and query tuning so a smaller instance suffices. Since there is no per-I/O charge, compute sizing is the main control.

Does C3X estimate AlloyDB cost?

C3X prices the AlloyDB cluster and instance resources, including compute configuration, from your Terraform, so compute and storage cost appear before deploy. Read-pool sizing depends on read load, which you assess from usage.

What to do next

Compare AlloyDB and Cloud SQL before you choose. C3X reads your Terraform and prices your database and the rest of your stack against a live catalog. Start with the quickstart.