High Performance PostgreSQL for Rails

Backfilling Large Tables Without Downtime

Over time, your database structure will evolve to meet the needs of the application. This could be from new features or more efficient designs. You may introduce a new empty column that needs to be populated for rows that existed before it was added. The column would be NULL to start.

The process of filling in data for existing rows is called “backfilling.” As your table row counts increase, backfills take longer and can become very complicated and risky. One approach is to try and avoid them and provide application-level data in lieu of data from the database.

This can be a worthwhile trade-off to spare the development for some cases. In those cases, the backfill could be called optional when the application ...