azure-vmss-autoscale-cli

Mini Azure Project – Linux VM Scale Set with CPU-based Autoscaling (Azure CLI)

This mini-lab demonstrates how to deploy and operate an Azure Virtual Machine Scale Set (VMSS) using Azure CLI only, including CPU-based autoscaling (scale-out & scale-in). The project is part of my AZ-104 learning path and focuses on compute scalability, availability, and automation, without using the Azure Portal for deployment.

---

Goals

• Create a Resource Group
• Deploy a Linux VM Scale Set (Uniform) via Azure CLI
• Configure CPU-based autoscaling
• Trigger Scale-Out and Scale-In events
• Observe VMSS behavior in a real scenario

---

Prerequisites

• Azure Subscription
• Azure CLI or Azure Cloud Shell
• Existing SSH key (~/.ssh/id_rsa.pub)
• Basic understanding of:
• Virtual Machines
• Load Balancers
• Metrics & autoscaling concepts

---

Step 1 - Create Resource Group

az group create \
  --name rg-vmss-autoscale \
  --location westeurope \
  --output table

---

Step 2 - Deploy Linux VM Scale Set (Uniform)

az vmss create \
  --resource-group rg-vmss-autoscale \
  --name vmss-linux-autoscale \
  --image Ubuntu2204 \
  --vm-sku Standard_B2s \
  --instance-count 1 \
  --authentication-type ssh \
  --admin-username "enteryouruserhere" \
  --ssh-key-values ~/.ssh/id_rsa.pub \
  --orchestration-mode Uniform \
  --upgrade-policy-mode automatic \
  --output table

Result:

• VMSS created with 1 instance
• Standard Load Balancer, VNet, Subnet and NSG are created automatically

Proof - VM Scale Set created

---

Step 3 - Configure Autoscaling (CPU-based)

Create Autoscale Profile

az monitor autoscale create \
  --resource-group rg-vmss-autoscale \
  --resource vmss-linux-autoscale \
  --resource-type Microsoft.Compute/virtualMachineScaleSets \
  --name autoscale-vmss-cpu \
  --min-count 1 \
  --max-count 3 \
  --count 1

Scale-Out Rule (CPU > 70%)

az monitor autoscale rule create \
  --resource-group rg-vmss-autoscale \
  --autoscale-name autoscale-vmss-cpu \
  --condition "Percentage CPU > 70 avg 5m" \
  --scale out 1

Scale-In Rule (CPU < 50%)

az monitor autoscale rule create \
  --resource-group rg-vmss-autoscale \
  --autoscale-name autoscale-vmss-cpu \
  --condition "Percentage CPU < 50 avg 10m" \
  --scale in 1

---

Step 4 - Observe VMSS Scaling Behavior

List VMMS Instances

az vmss list-instances \
  --resource-group rg-vmss-autoscale \
  --name vmss-linux-autoscale \
  --query "[].instanceId" \
  --output table

Observed behavior

• Initial state: 1 instance
• Under load: scaled out to 2 and 3 instances
• After load stopped: scaled back down to 2, then 1
• Scale-In takes longer due to averaging windows and cooldown behavior.

Proof - Scale-Out and Scale-in

Initial State:

Scale-Out (CPU > 70%):

Scale-In (CPU < 50%):

---

Section 5 - Cleanup

az group delete \
  --name rg-vmss-autoscale \
  --yes --no-wait

---

Learnings

• VM Scale Sets provide horizontal scalability for virtual machines
• Autoscale decisions are not instant – scale-in is intentionally slower than scale-out
• Autoscaling is driven by:
• Metric thresholds
• Averaging windows
• Cooldown periods
• VMSS abstracts:
• Load Balancer integration
• Instance lifecycle management
• CLI-only deployments expose real operational details often hidden in the Portal
• VMSS is a foundation service used before higher-level services like App Services or containers

---

Notes

This project focuses on infrastructure-level scaling, not application deployment

VMSS is commonly used as a building block for:

• High-availability compute
• Backend services
• Autoscaled workloads