Most engineers read a Terraform plan like a code diff.
That mental model hides structural risk.
Terraform does not execute configuration top to bottom. It compiles it into a directed graph. Resources become nodes. References become edges. Execution order is determined by data flow, not file position.
This matters because identity is not just “same settings.” Identity is the combination of the resource address in state and the remote object ID from the provider. Change the address and Terraform may destroy and recreate, even if your intent did not change.
In this lesson, I break down:
- How implicit dependencies are created
- Why unknown values split execution phases
- When explicit dependencies reduce parallelism
- How to review a plan by tracing structural impact instead of diff noise
If you work with Terraform in production, this mental model reduces surprise during refactors.