连接Cloud SQL的BigQuery计费咨询：存储成本及部分表访问计费方式

Hey there! Let's walk through your cost questions step by step—connecting BigQuery to Cloud SQL via federated queries has a few nuanced billing points, so I’ll break it down plainly.

Do I need to pay storage costs in this scenario? #

Yes, but it depends on which service’s storage we’re talking about:

• Cloud SQL Storage: You’ll always pay for the storage used by your Cloud SQL database, regardless of whether you access it via BigQuery or not. This follows Cloud SQL’s standard pricing (based on storage type, region, and total data stored).
• BigQuery Storage: You only pay BigQuery storage costs if you persist data in BigQuery itself. If you’re just running federated queries directly against Cloud SQL (and not saving results to a BigQuery table/view), BigQuery won’t charge you for storage. But if you create a materialized view, save query results to a BigQuery table, or import Cloud SQL data into BigQuery, that data stored in BigQuery will incur storage fees.

How does billing work when accessing only 3 out of 10 Cloud SQL tables? #

For federated queries (the way BigQuery accesses Cloud SQL data directly), billing is based on data processed/scanned, not storage. Here’s the breakdown:

• When you run a query that only touches 3 of your 10 Cloud SQL tables, BigQuery will calculate the total amount of data it reads from those 3 tables during the query. This is what you’ll be charged for (under BigQuery’s "external data query" pricing category).
• If your query uses filters (like WHERE clauses) to limit the data read (e.g., only fetching the last 30 days of records), you’ll only be billed for the filtered data that’s actually scanned—not the full size of the 3 tables.
• Remember: Data Studio itself doesn’t add extra storage or query costs—it just visualizes the results of your BigQuery queries. All costs come from Cloud SQL’s storage/resource fees and BigQuery’s federated query fees (plus any BigQuery storage if you persist data there).

Quick Example #

Suppose your 3 target Cloud SQL tables total 15GB, but your query only scans 2GB of data (thanks to a WHERE date >= '2024-01-01' filter):

• Cloud SQL still charges you for the full 10 tables’ storage (since all data is stored there).
• BigQuery charges you for the 2GB of data scanned via the federated query.
• If you save that 2GB query result into a BigQuery table (say, it compresses to 500MB), you’ll pay monthly BigQuery storage fees for that 500MB until you delete the table.

内容的提问来源于stack exchange，提问作者thienpham234