What a SELECT * view costs in BigQuery

What it is

A logical view is expanded into the query that references it, and BigQuery bills the columns the top-level query uses, directly or through the view. So SELECT user_id FROM a SELECT * view bills user_id alone. The cost arrives when the reader selects everything too: SELECT * FROM the view, or a BI tool that pulls every field.

Why it costs money

• Readers tend to treat a view as a table and select all of it, BI tools and extracts especially. Then every column of the underlying table is read, including ones the view never needed to expose.
• A SELECT * view changes shape whenever the table does: a new large JSON column appears in the view, and in every SELECT * over it, with no change to anyone's SQL.
• Nested views hide what a query reads. Nobody reads three levels of definitions before writing a query.

Worked example

CREATE VIEW `analytics.v_events` AS
SELECT * FROM `analytics.events_raw`

CREATE VIEW `analytics.v_events` AS
SELECT event_id, user_id, event_name, event_timestamp, event_date
FROM `analytics.events_raw`

How to fix it

1. 1Enumerate columns in view definitions: the view keeps a stable shape, and what it can cost is clear from its definition.
2. 2Point BI tools and extracts at views that expose only the columns they need, since they tend to select everything.
3. 3Audit view definitions with INFORMATION_SCHEMA.VIEWS and search for SELECT *.
4. 4Prefer authorized views with an explicit column list where the view exists for access control.