- Tags:: đź“„Tech_docs
- Author:: Gitlab
- Link:: dbt Guide | GitLab
- Source date:: Unknown
- Finished date:: 2021-05-01
Ideally, source models should cast every column. Explicit is better than implicit. Test your assumptions.
Golden Data Tests: Some data is so important it should always exist in your data warehouse and should never change: your top customer’s record, the 2019 total global users count, the KPI result when you passed 1,000,000 subscriptions. Some of these cases can be solved by a developing new database capabilities, but this can be complicated and may not always match your existing data processing workflow. In addition, bugs can be accidentally added to data transforms or someone can accidentally run a bad UPDATE versus a critical production table. The Golden Data test is a specialized type of Column Value test that validates the existence of a known data literal and helps catch these problems when they occur