Question

A retail application runs on GKE and uses a Cloud SQL for PostgreSQL instance. During peak sales events, the product catalog endpoint experiences high latency and the Cloud SQL CPU is consistently pinned at 95%. Analysis shows that 80% of the database queries are identical reads for popular product details that change infrequently. The team wants to reduce database load and improve read latency without increasing the Cloud SQL machine size. What should they implement?

Accepted Answer

Add a Memorystore for Redis cache layer in front of Cloud SQL and serve frequently requested product reads from the cache with an appropriate TTL.

Answer

Enable Cloud SQL storage auto-resize so the instance can grow its disk to handle the increased query volume.

Answer

Increase the GKE Horizontal Pod Autoscaler maximum replicas so more application pods can send queries to Cloud SQL in parallel.

Answer

Switch the Cloud SQL instance to a higher persistent disk IOPS tier to speed up the repeated read queries.