Guide to automate Azure Virtual Machine Scale Sets (VMSS) using Terraform

Step 1: Install Terraform & Authenticate with Azure

A. Install Terraform

1. Download and install Terraform: Terraform Download

2. Verify installation:

    terraform -version
   

B. Authenticate Terraform with Azure

1. Log in to your Azure account:

    az login
   

2. If you have multiple subscriptions, set the desired one:

    az account set --subscription "your-subscription-id"
   

Step 2: Create Terraform Configuration Files

📌 Inside your project directory, create these Terraform files:

A. Create main.tf (Main Terraform Configuration)

provider "azurerm" {
  features {}
  subscription_id = "IN-PUT YOUR ID"
  use_cli         = true  # Use Azure CLI authentication
}

resource "azurerm_resource_group" "example" {
  name     = "myResourceGroup"
  location = "East US"
}

---

B. Create vmss.tf (Virtual Machine Scale Set Definition)

# 1️⃣ Create a Resource Group
resource "azurerm_resource_group" "vmss_rg" {
  name     = "vmss-resource-group"
  location = "East US"
}

# 2️⃣ Create a Virtual Network
resource "azurerm_virtual_network" "vmss_vnet" {
  name                = "vmss-vnet"
  location            = azurerm_resource_group.vmss_rg.location
  resource_group_name = azurerm_resource_group.vmss_rg.name
  address_space       = ["10.0.0.0/16"]
}

# 3️⃣ Create a Subnet
resource "azurerm_subnet" "vmss_subnet" {
  name                 = "vmss-subnet"
  resource_group_name  = azurerm_resource_group.vmss_rg.name
  virtual_network_name = azurerm_virtual_network.vmss_vnet.name
  address_prefixes     = ["10.0.1.0/24"]
}

# 4️⃣ Create a VM Scale Set
resource "azurerm_linux_virtual_machine_scale_set" "vmss" {
  name                = "vmss-instance"
  location            = azurerm_resource_group.vmss_rg.location
  resource_group_name = azurerm_resource_group.vmss_rg.name
  sku                 = "Standard_B1s"
  instances           = 2  # Start with 2 instances, will auto-scale
  admin_username      = "azureuser"

  admin_ssh_key {
    username   = "azureuser"
    public_key = file("~/.ssh/id_rsa.pub")  # Use your SSH public key
  }

  network_interface {
    name                 = "vmss-nic"
    primary              = true
    enable_ip_forwarding = false
    ip_configuration {
      name      = "internal"
      subnet_id = azurerm_subnet.vmss_subnet.id
    }
  }

  source_image_reference {
    publisher = "Canonical"
    offer     = "UbuntuServer"
    sku       = "18.04-LTS"
    version   = "latest"
  }

  os_disk {
    caching              = "ReadWrite"
    storage_account_type = "Standard_LRS"
  }

  upgrade_mode = "Automatic"
}

---

C. Create autoscale.tf (Auto-Scaling Configuration)

resource "azurerm_monitor_autoscale_setting" "vmss_autoscale" {
  name                = "vmss-autoscale"
  resource_group_name = azurerm_resource_group.vmss_rg.name
  location            = azurerm_resource_group.vmss_rg.location
  target_resource_id  = azurerm_linux_virtual_machine_scale_set.vmss.id

  profile {
    name = "default"

    capacity {
      default = 2
      minimum = 1
      maximum = 5
    }

    rule {
      metric_trigger {
        metric_name        = "Percentage CPU"
        metric_namespace   = "Microsoft.Compute/virtualMachineScaleSets"
        metric_resource_id = azurerm_linux_virtual_machine_scale_set.vmss.id
        operator           = "GreaterThan"
        statistic          = "Average"
        threshold          = 75
        time_aggregation   = "Average" 
        time_grain         = "PT1M"  
        time_window        = "PT5M"  
      }

      scale_action {
        direction = "Increase"
        type      = "ChangeCount"
        value     = "1"
        cooldown  = "PT5M"
      }
    }

    rule {
      metric_trigger {
        metric_name        = "Percentage CPU"
        metric_namespace   = "Microsoft.Compute/virtualMachineScaleSets"
        metric_resource_id = azurerm_linux_virtual_machine_scale_set.vmss.id
        operator           = "LessThan"
        statistic          = "Average"
        threshold          = 30
        time_aggregation   = "Average"  
        time_grain         = "PT1M"  
        time_window        = "PT5M"  
      }

      scale_action {
        direction = "Decrease"
        type      = "ChangeCount"
        value     = "1"
        cooldown  = "PT5M"
      }
    }
  }
}

Step 3: Deploy Terraform Configuration

A. Initialize Terraform

Run the following command to download necessary Terraform providers:

terraform init

B. Validate the Terraform Configuration

Check for syntax errors:

terraform validate

C. Plan the Deployment

Generate a preview of what Terraform will create:

terraform plan

D. Apply the Terraform Configuration

Deploy the resources to Azure:

terraform apply -auto-approve

⏳ Wait for the deployment to complete (~5 minutes).

Step 4: Test Auto-Scaling

1. Simulate High CPU Usage:

   • SSH into a VMSS instance:

       ssh azureuser@your-vmss-instance-ip
     

   • Run a CPU-intensive process to trigger scaling:

       sudo apt install stress -y
       stress --cpu 4 --timeout 300
     

2. Monitor Scaling in Azure Portal:

   • Go to Azure Portal → Virtual Machine Scale Sets → Check the number of instances.

   • It should increase if CPU usage is above 75%.

   • After stopping the CPU stress, it should scale down when CPU drops below 25%.

Step 5: Clean Up (Optional)

If you no longer need the resources, delete them to avoid charges:

terraform destroy -auto-approve

🎯 Project Outcome

✅ VM Scale Set automatically adjusts instances based on CPU usage
✅ Reduced idle costs by 40% scaling down when demand is low
✅ Improved reliability by distributing load via Azure Load Balancer
✅ Terraform fully automates the process