Automating AI Infrastructure on GCP with Infrastructure as Code.md

Automating AI Infrastructure on GCP with Infrastructure as Code

1. Open Cloud Shell
2. Enable the required APIs

gcloud services enable compute.googleapis.com \
                         iap.googleapis.com \
                         aiplatform.googleapis.com \
                         notebooks.googleapis.com \
                         cloudkms.googleapis.com \
                         cloudresourcemanager.googleapis.com \
                         iam.googleapis.com

3. Initiate the folders

mkdir tf-vertex-ai-[name]
cd tf-vertex-ai-[name]

4. Create the prerequisite files

touch main.tf network.tf outputs.tf variables.tf

5. Edit main.tf

resource "google_service_account" "workbench_sa" {
  account_id   = "vertex-ai-workbench-sa-[name]"
  display_name = "Vertex AI Workbench Service Account"
  project      = var.project_id
}

module "simple_vertex_ai_workbench" {
  source  = "GoogleCloudPlatform/vertex-ai/google//modules/workbench"
  version = "~> 0.2"

  name       = "simple-vertex-ai-workbench-[name]"
  location   = "us-central1-a"
  project_id = var.project_id

  machine_type = "e2-standard-2"

  labels = {
    env  = "test"
    type = "workbench"
  }

  service_accounts = [
    {
      email = google_service_account.workbench_sa.email
    },
  ]

  data_disks = [
    {
      disk_size_gb = 20
      disk_type    = "PD_BALANCED"
    },
  ]

  network_interfaces = [
    {
      network  = module.test-vpc-module.network_id
      subnet   = module.test-vpc-module.subnets_ids[0]
      nic_type = "GVNIC"
    }
  ]
}

6. Edit network.tf

resource "random_id" "suffix" {
  byte_length = 4
}

module "test-vpc-module" {
  source       = "terraform-google-modules/network/google"
  version      = "~> 9.0"
  project_id   = var.project_id
  network_name = "simple-workbench-[name]"
  mtu          = 1460
  subnets = [
    {
      subnet_name           = "simple-subnet-01-[name]"
      subnet_ip             = "10.10.1X.0/24"
      subnet_region         = "us-central1"
      subnet_private_access = true
    },
  ]
}

7. Edit variables.tf

variable "project_id" {
  description = "The ID of the project in which the resource belongs"
  type        = string
}

8. Edit outputs.tf

output "project_id" {
  value       = var.project_id
  description = "The project ID"
}

output "workbench_name" {
  value       = module.simple_vertex_ai_workbench.workbench_name
  description = "The name of the Vertex AI Workbench instance"
}

output "workbench" {
  value       = module.simple_vertex_ai_workbench
  description = "The Vertex AI Workbench instance"
}

output "service_account" {
  value       = google_service_account.workbench_sa.email
  description = "The service account for Vertex AI Workbench instance"
}

output "network" {
  value       = module.test-vpc-module.network_self_link
  description = "The network for Vertex AI Workbench instance"
}

output "subnet" {
  value       = module.test-vpc-module.subnets_self_links[0]
  description = "The subnet for Vertex AI Workbench instance"
}

9. Export the project_id

export TF_VAR_project_id="your_project_id"

10. Init the terraform project

terraform init

11. Plan the Terraform project

terraform plan

12. Apply the Terraform project

terraform apply

13. Destroy the Terraform project

terraform destroy