Introduction

Ask a team how their production servers were set up, and often the honest answer is "someone clicked through the console two years ago". The security group has rules nobody remembers adding. The test environment is "almost the same" as production. When a region goes down, rebuilding everything elsewhere would take days of guesswork.

Infrastructure as code (IaC) is the answer to that. Instead of clicking, you describe the infrastructure you want in text files. A tool reads the files and makes reality match them. The files go into Git, so every change has an author, a date and a review, and you can build the same environment again whenever you need it.

Terraform, made by HashiCorp, is the most widely used IaC tool for this job.

What Terraform does

You write .tf files in HCL (HashiCorp Configuration Language) that say what you want:

resource "aws_instance" "web" {
  ami           = "ami-xxxxxxxxxxxxxxxxx"
  instance_type = "t3.micro"
}

Then you run two commands:

terraform plan    # show me what you would change
terraform apply   # make it so

Terraform compares what you asked for with what already exists, shows you the difference, and on your approval calls the AWS APIs to create, update or delete exactly what is needed. Run it again with no changes and it does nothing. Change one line and it changes one thing.

This style is called declarative: you describe the end state, not the steps. You never write "call RunInstances, wait, then tag it". Terraform works out the order itself, including dependencies such as "create the security group before the instance that uses it".

Core vocabulary

You will meet each of these in the course, roughly in this order.

Why not just use CloudFormation?

AWS has its own IaC service, CloudFormation, and Azure and Google Cloud have theirs. They work well, but each one only speaks to its own cloud. Terraform uses the same language and the same workflow for every platform that has a provider: AWS, Azure, Google Cloud, Kubernetes, GitHub, Cloudflare, Datadog and thousands more. Learn it once, use it everywhere.

Terraform also gives you the plan step as a first class part of the workflow, and a huge public registry of providers and community modules.

Where Terraform fits with other DevOps tools

Terraform's job is provisioning: creating the infrastructure. It is not the best tool for installing and configuring software inside servers, which is configuration management, the job of tools like Ansible.

A typical flow in a company: an engineer changes a .tf file and opens a pull request, the pipeline runs terraform plan and posts the result, a reviewer approves, and the pipeline runs terraform apply. Nobody touches the console.

Things to know before you start

• State is precious. Terraform's memory of what it created lives in the state file. Lose it, and Terraform loses track of your infrastructure. The course shows you what that looks like and how teams protect it.
• Credentials stay out of code. Terraform reads AWS credentials from the environment, a profile or an IAM role. They never belong in .tf files.
• Always read the plan. terraform apply will happily delete a database if your code says so. The plan is where you catch that.
• Clean up. terraform destroy removes everything a configuration created. Use it at the end of every practice session.

Start with Why Terraform and What is Terraform and work through the sidebar in order. Each lesson builds on the files from the one before.

Written By: Shramik Masti