Advanced|18 hours|35 lessons

Terraform in Production

Operating a Terraform estate as a production system: state as a database with locking and corruption modes, blast radius and the arithmetic that sets the state boundary, drift and the plan that does not match the apply, refactoring live infrastructure, modules across many teams, and who is allowed to apply. Lifetime updates included.

Text-based, no videos

9 modules, 35 lessons

Lifetime access

What you'll learn

The state file as a system you operate, with locking, corruption modes and a recovery path

How the state boundary sets blast radius, and the plan time and lock contention arithmetic behind it

Detecting drift, deciding what to do about it, and which drift to accept

Why the apply sometimes does not match the plan, and how to work that failure

Refactoring live infrastructure: renames, imports, and moving resources between states

Designing module interfaces that many teams can adopt, and migrating consumers without a freeze

What infrastructure code can actually be tested, and what only an apply will tell you

Account structure, the apply identity, and running the estate for many teams

Curriculum

9 modules · 35 lessons

State Is a Database You Operate

The state file as a system with consistency, locking and corruption modes, rather than a file you never look at.

4 lessons

What the State File Actually Is24 minFREE Locking, and What Happens Without It30 minFREE Corruption, Recovery and the Backup You Do Not Have28 minFREE Reading a Plan Correctly26 minFREE

Blast Radius and the State Boundary

How you split state decides what one mistake can destroy, and the arithmetic that sets the boundary.

4 lessons

What One Apply Can Destroy28 min The Arithmetic of Splitting State35 min Splitting State Without an Outage32 min Cross State Coupling30 min

The Plan Is a Prediction

What the plan is actually claiming, and the several ways reality declines to match it.

5 lessons

Refresh, and Why the Plan Changed30 min Drift and What Changed Outside Terraform35 min When the Apply Does Not Match the Plan32 min Provider Behaviour and the Non Declarative Corners33 min Replacement You Did Not Ask For28 min

Dependencies and Apply Order

The graph Terraform builds, the edges it cannot see, and what a partial apply leaves behind.

4 lessons

The Graph Terraform Builds28 min The Dependency You Did Not Declare32 min Partial Applies and Half Built Environments30 min Create Before Destroy27 min

Modules at Organisation Scale

Sharing infrastructure definitions across many teams without a registry full of things nobody can change.

4 lessons

What Belongs in a Module26 min Versioning and Consumer Drift30 min Interface Design and the Parameterisation Trap28 min Migrating Consumers Without a Freeze30 min

Refactoring Live Infrastructure

Changing the shape of the code without changing what is running, which the tool does not do for you.

4 lessons

Renaming Destroys30 min Importing What Already Exists32 min Moving Resources Between States33 min Deleting Safely28 min

Testing and Policy

What can be verified before an apply, and what can only be verified by applying.

4 lessons

What You Can Actually Test28 min Plan Based Testing and Policy as Code32 min Ephemeral Environments30 min Reviewing a Plan You Did Not Write27 min

Running It for Many Teams

The platform side: bootstrapping, account structure, who is allowed to apply, and staying current.

5 lessons

The Bootstrap Problem26 min Multi Account and Multi Region Structure32 min The Apply Identity30 min The IaC Pipeline28 min Version and Provider Upgrades30 min

Capstone

A complete Terraform estate design combining every module, with explicit tradeoffs.

1 lessons

Capstone: Design the Terraform Estate60 min

About the Author

Sharon Sahadevan

AI Infrastructure Engineer

Building production GPU clusters on Kubernetes. H100s, large-scale model serving, and end-to-end ML infrastructure across Azure and AWS.

10+ years designing cloud-native platforms with deep expertise in Kubernetes orchestration, GitOps (Argo CD), Terraform, and MLOps pipelines for LLM deployment.

Author of KubeNatives, a weekly newsletter read by 3,000+ DevOps and ML engineers for production insights on K8s internals, GPU scheduling, and model-serving patterns.

Ready to master this topic?

Start with the free preview lesson and see for yourself.