Hacker News Reader

For ad hoc copies I’ve found the third party schema cloning function useful. I assume performance would be an issue for huge databases though. https://github.com/denishpatel/pg-clone-schema

View on HN

The difference is actually copying the data vs. copy-on-write. With the latter, you can make infinite copies without worrying about how much space you're using or the time it takes to make a copy, thus making it feasible to start with a fresh copy of the prod DB before you do literally anything. Have CI/CD create a fresh copy before every test. Have every feature branch start with a clean copy of prod and give people a button to do it again if they mess up a migration.

COW unlocks a ton of interesting use cases. It's easy to do on AWS Aurora but unfortunately their minimum instance size is way bigger than the regular RDS instances.

Reply on news.ycombinator.com