Re:Staging is dead: The rise of preview environments
by jcraft on Thursday December 22, 2022 @01:44PM
(#34096831)
There are ways to anonymize production data and make it available in preview and testing environments. My company has been doing that for customers and it has been effective. You get privacy-safe data that is still very close to production data.
Depending on the application architecture, you can spin up database containers and seed them with replicas of the anonymized data.
The usual staging and/or preproduction environments would be the preview of their own branches.
It sounds nice. The only problems I can see are:
1) The cost, because each one of those instances could have its own queues, databases, connections with third party APIs, etc.
2) Seeding the database with a meaningful amount of data. I often see preproduction database built by testers month by month after a minimal seeding. Each instance here has to be seeded with all data plus what's required to demo the new feature. That must be saved before destroying the database, reused for further branches, merged with the data for other features.