Getting started with Terraform on Google Cloud

This chapter follows Infrastructure as Code with Terraform. You need a GCP project with billing enabled, the Google Cloud SDK (gcloud), and Terraform 1.x installed.

Prepare the workstation

Install Terraform from developer.hashicorp.com/terraform/downloads or your package manager. Confirm the binary:

1
terraform version

1
gcloud auth application-default login
2
gcloud config set project YOUR_PROJECT_ID

Enable APIs Terraform will call (one-time per project):

1
gcloud services enable storage.googleapis.com compute.googleapis.com

Create a working directory for this tutorial, for example ~/terraform-gcp-intro.

Configure providers and backend (local state)

Create versions.tf to pin Terraform and the Google provider:

1
terraform {
2
  required_version = ">= 1.5.0"
3

4
  required_providers {
5
    google = {
6
      source  = "hashicorp/google"
7
      version = "~> 5.0"
8
    }
9
  }
10
}
11

12
provider "google" {
13
  project = var.project_id
14
  region  = var.region
15
}

For learning, state stays local in terraform.tfstate. Before team use, move state to a GCS backend as described in State data in Terraform.

Add variables.tf:

1
variable "project_id" {
2
  type        = string
3
  description = "GCP project ID"
4
}
5

6
variable "region" {
7
  type        = string
8
  default     = "europe-west1"
9
  description = "Default region for regional resources"
10
}
11

12
variable "bucket_name" {
13
  type        = string
14
  description = "Globally unique GCS bucket name"
15
}

Create terraform.tfvars (do not commit secrets here; project id and bucket name are fine):

1
project_id  = "YOUR_PROJECT_ID"
2
bucket_name = "YOUR_UNIQUE_BUCKET_NAME"

Create a Cloud Storage bucket

Add main.tf with a single bucket:

1
resource "google_storage_bucket" "tutorial" {
2
  name     = var.bucket_name
3
  location = var.region
4

5
  uniform_bucket_level_access = true
6

7
  force_destroy = true
8
}

Add outputs.tf:

1
output "bucket_url" {
2
  value       = google_storage_bucket.tutorial.url
3
  description = "Base URL of the tutorial bucket"
4
}

Initialize and apply:

1
cd ~/terraform-gcp-intro
2
terraform init
3
terraform fmt -recursive
4
terraform validate
5
terraform plan
6
terraform apply

Open the bucket URL from terraform output bucket_url or verify in the Cloud Console. State now records the bucket resource id.

Add a VPC and a small VM

Extend main.tf with networking and a minimal instance:

1
resource "google_storage_bucket" "tutorial" {
2
  name     = var.bucket_name
3
  location = var.region
4

5
  uniform_bucket_level_access = true
6
  force_destroy               = true
7
}
8

9
resource "google_compute_network" "vpc" {
10
  name                    = "terraform-tutorial-vpc"
11
  auto_create_subnetworks = false
12
}
13

14
resource "google_compute_subnetwork" "subnet" {
15
  name          = "terraform-tutorial-subnet"
16
  ip_cidr_range = "10.0.1.0/24"
17
  region        = var.region
18
  network       = google_compute_network.vpc.id
19
}
20

21
resource "google_compute_instance" "vm" {
22
  name         = "terraform-tutorial-vm"
23
  machine_type = "e2-micro"
24
  zone         = "${var.region}-b"
25

26
  boot_disk {
27
    initialize_params {
28
      image = "debian-cloud/debian-12"
29
    }
30
  }
31

32
  network_interface {
33
    network    = google_compute_network.vpc.id
34
    subnetwork = google_compute_subnetwork.subnet.id
35
  }
36

37
  tags = ["terraform-tutorial"]
38
}

Run plan again. Terraform should propose the network, subnet, and VM without recreating the bucket unless you changed it:

1
terraform plan
2
terraform apply

Inspect resources in the console or with gcloud compute instances list.

Clean up

When you are finished experimenting, destroy managed resources:

1
terraform destroy

Confirm when prompted. force_destroy = true on the bucket allows deletion even if objects exist—appropriate for a sandbox, not for production data stores.

Next: HCL structure in Terraform explains blocks, expressions, and meta-arguments in depth.