New: Cookbooks and AI ExplanationsStep-by-Step recipes to solve problems connected to Roadmaps and Cheat Sheets. Need more details? Use AI buttons for structured and simple explanations with concrete examples throughout the whole platform.Take a look
Testing a change against real data means either risking production or working from a stale, sampled backup.
What you'll have at the end
A cloned database that mirrors production exactly, with storage cost confirmed at zero until you write to it
You need
A production Snowflake database you already have read access to, and a role permitted to create new databases in the account.
Not covered
Undoing a mistake already committed in production.
Give dbt a dev target
when the tool doing the testing is dbt itself and you want its own unlabeled commands defaulting somewhere safe on every run, permanently
Is a resource monitor stopping your next runaway Snowflake bill?
when the sandbox's own compute, not its storage, is the cost you're worried about
Undo a bad update with Time Travel
when the mistake already happened in production itself, not in a sandbox you're about to test something new in
Checked 26 Aug 2026
Part of the Snowflake Data Cloud cookbook