Cloud SQL Right-sizing (GCP)
CostLens surfaces Cloud SQL over-provisioned recommendations so you can resize instances that are larger than their workload needs.
CostLens surfaces over-provisioned Cloud SQL instances flagged by Google's Overprovisioned Recommender — instances running a larger machine tier than their actual workload requires. Over-provisioned databases pay for vCPU and memory they never use.
How it works
Recommender analyses utilisation
Google's Cloud SQL Overprovisioned Recommender evaluates CPU and memory usage per instance and suggests a smaller tier.
CostLens reads them read-only
Using the recommender.viewer scope from GCP Accounts, CostLens ingests each recommendation and its projected saving.
Severity and savings are assigned
Savings come from the Recommender's costProjection, normalised to a 30-day month.
What to do
- Review the recommended tier and confirm it meets your peak requirements (headroom for spikes).
- Resize the instance in the Cloud Console (Cloud SQL → Edit → Machine type).
Resize causes a brief restart
Changing a Cloud SQL machine type restarts the instance. Schedule it during a maintenance window and confirm your application reconnects cleanly.
Severity levels
| Severity | Estimated monthly savings |
|---|---|
| critical | >$500 |
| high | $100–$500 |
| medium | $20–$100 |
| low | <$20 |
Required permissions
- Detect:
roles/recommender.viewer— see GCP Accounts.
Limitations
- Recommendations are based on Google's observation window; instances with variable or seasonal load may need judgement before resizing.
- This check is advisory — CostLens surfaces the recommendation but the resize is applied in the Cloud Console.