Terraform can feel unpredictable at first.
Resources create in parallel.
Updates ripple unexpectedly.
A small change suddenly forces replacement.
None of that is random.
Terraform does not execute files.
It executes a plan, and that plan is built from a dependency graph.
Every reference you write forms a relationship.
Every relationship becomes an edge.
Together, they form a Directed Acyclic Graph that controls when things are created, what must update, and why destruction happens in reverse order.
This is also where many misunderstandings begin.
The graph determines order, but it is the tracked identity in state that tells Terraform which real object to act on. Without identity, Terraform cannot safely update or destroy anything. It can only guess and guessing leads to replacement.
In this lesson, you will see:
- Why Terraform creates resources in waves
- How updates propagate through dependencies
- Why some changes cannot be mutated and must be replaced
- How destroy is simply the graph, reversed
- Why state is memory, not an implementation detail
If Terraform has ever surprised you, this lesson explains why.
Watch the full video below to build the correct mental model before moving deeper into Terraform.