Question

A team runs a self-managed PostgreSQL database on a Compute Engine n2-standard-16 instance backed by a 500 GB balanced persistent disk (pd-balanced). During peak transaction load, the application experiences high query latency, and monitoring shows the disk consistently hitting its IOPS and throughput limits while CPU and memory utilization remain under 40%. The team wants to improve database performance with the least operational disruption and without over-provisioning compute. What should they do first?

Accepted Answer

Increase the persistent disk size and/or switch to pd-ssd to raise the IOPS and throughput limits, since these scale with disk type and provisioned capacity.

Answer

Migrate the workload to a larger machine type such as n2-standard-32 to raise the per-VM disk performance ceiling and IOPS baseline.

Answer

Enable a Cloud CDN in front of the database to cache frequent query results and reduce disk reads.

Answer

Reduce the disk size to force the autoscaler to provision additional read replicas automatically.