Different environments are just a trick to meet various stages of production life cycle - development, testing/staging and release into production. The number of stages can vary from project to project depending on complexity and habits, but the classical is a 3-stage pipeline. The main things here are isolation and duplication: all stages must be isolated from each other and the staging environment must duplicate a production environment. ==You should be ready for unexpected consequences if principles are not satisfied==. It's common practice for teams and studios to roll out staging instances with data from the production database to test recent updates.
!!! abstract "All stages have their own instances of every project entity" - Client build; - Backend instance; - Developer account; - Url; - Analytics account; - And any other things.
So you should have 1 separated Goblin Backend instance for every stage with its own cluster and database.
The development environment instance is free for onboarding, has minimum power and workloads also don't represent a cluster. The staging environment instance is represented as cluster but has minimum workloads so you can pick S1-h1 pack. So and production environment instance will have the maximum workload - you should pick the right pack based on your DAU and CCU needs.