Skip to main content
Sponsor @cloudposse

Decide on Seeding Strategy for Staging Environments

Problem​

Longer-lived staging environments need a dataset that closely resembles production. If this dataset becomes stale, we’ll not be effectively testing releases before they hit production. Restoring snapshots from production is not recommended.

Considerations​

  • Should contain anonymized users, invalid email addresses

  • No CHD, PHI, PII must be contained in the database

  • The scale of data should be close to the production database

  • Snapshots from production are dangerous if not anonymized/scrubbed (imagine the risk of sending emails to everyone from your staging env)

  • Fixtures are not recommended (scale of data for fixtures usually does not represent production)

  • We recommend including the DBA in these conversations.

  • QA teams want stable data so that they can run through their test scenarios

Recommendations​

caution

Cloud Posse does not have a turnkey solution for seeding staging environments

  • ETL pipeline scrubs the data and refreshes the database weekly or monthly. (e.g. AWS Glue, GitHub Action Schedule Job)