Check Cloud SQL compute, storage, backup and network charges

Why this is worth a look

The detailed usage cost export includes standard billing data plus granular, resource-level cost information. Cloud SQL pricing depends on provisioned storage, selected CPUs, selected memory, and hosting location. The detailed-export reference also excludes some network data transfer, Cloud SQL backups in specific locations, and Storage Snapshot costs from granular Cloud SQL instance costs. Use documented monitoring metrics as usage context, not as billing rows.

A read-only manual procedure for reviewing one complete BigQuery detailed billing export period and comparing the reviewed billing information with documented Cloud SQL for MySQL metrics.

READ-ONLY MANUAL CHECKLIST

Inputs: the fully qualified BigQuery detailed usage cost export table selected by your organization; one invoice-month window; the Cloud SQL project or billing scope under review.

1. In BigQuery, open the detailed usage cost export table for the selected invoice month. Confirm that detailed usage cost export is enabled and that the period is present. Do not use the Standard usage export for this review.
2. Filter the table to rows whose displayed service is Cloud SQL, using the service value shown in the table. Review the exported billing descriptions, resource-level information, and cost values exactly as displayed. Do not treat these as fixed field names or invent a net-cost calculation.
3. Keep each reviewed row together with its displayed service, billing description, resource-level information, and cost value. Use an unclassified note for rows whose displayed information does not support a compute, storage, backup, or network label.
4. For Cloud SQL for MySQL only, open the Cloud SQL instance Overview page or Cloud Monitoring and review storage usage, CPU usage, and memory usage for the same invoice-month window. Use those metrics to explain usage patterns, not to replace billing rows.
5. Check export completeness and timing. Note a first-time backfill, a period when export was disabled, or recent usage whose costs are not yet available. Keep the documented exclusions separate from granular Cloud SQL instance costs.

Execution surface: BigQuery and the Cloud SQL console or Cloud Monitoring. Required permissions are read access arranged by your organization to the billing-export dataset and the relevant Cloud SQL monitoring view. Preserve the cost units displayed by the export and use one invoice-month window. These steps make no changes to billing, instances, backups, or monitoring.

How to confirm it

1. 01

   Choose a complete invoice month

   Open the detailed usage cost export in BigQuery and select one invoice month. Confirm that the period is present and that detailed export is enabled. If it was recently enabled, allow up to five days for the documented initial retroactive backfill to finish.

2. 02

   Review displayed billing information

   Filter to Cloud SQL rows using the service value displayed in the table. Keep each row's displayed billing description, resource-level information, and cost value together. These are descriptions of what the export shows, not assumed column names or a derived net-cost measure.

3. 03

   Keep unclear categories unclassified

   Label a row as compute, storage, backup, or network only when its displayed billing information supports that interpretation. Otherwise record it as unclassified for follow-up. Do not calculate net cost from fields that have not been verified.

4. 04

   Compare MySQL monitoring metrics

   For Cloud SQL for MySQL only, use the instance Overview page or Cloud Monitoring to compare storage usage, CPU usage, and memory usage with the invoice-month window. Use the metrics to explain usage patterns, not as billing records.

5. 05

   Check timing and exclusions

   Account for delays before recent costs appear and for periods when detailed export was disabled. Keep some network data transfer, Cloud SQL backups in specific locations, and Storage Snapshot costs separate from granular Cloud SQL instance costs.

Primary sources