Skip to main content

04. AWS ASG

An Auto Scaling Group (ASG) in AWS automatically manages the number of EC2 instances based on application load and health. It ensures that we always have the right number of instances running to handle current traffic efficiently.

🧩 Why Use Auto Scaling Groups?​

Horizontal scaling = scale out / scale in

BenefitDescription
⚙️ ElasticityScale out (add instances) or scale in (remove instances) automatically to match demand.
💰 Cost OptimizationPay only for the resources you need — no overprovisioning.
❤️ High AvailabilityAutomatically replaces unhealthy instances with new healthy ones.
🔄 Integration with Load BalancerASG works hand-in-hand with an Application Load Balancer (ALB) to distribute traffic evenly.

⚙️ How Auto Scaling Works​

  1. Define Group Parameters

    • Minimum size: Lowest number of EC2 instances always running.
    • Desired capacity: The target number of instances currently running.
    • Maximum size: The upper limit of instances allowed in the group.

  2. Scale Out

    • Automatically adds EC2 instances when load increases (e.g., more traffic during the day).

  3. Scale In

    • Automatically removes EC2 instances when load decreases (e.g., less traffic at night).

  4. Health Check & Replacement

    • ASG automatically terminates unhealthy instances and launches new ones.

🌐 Load Balancer Integration​

  • ASG instances are automatically registered and deregistered from the Load Balancer.
  • As instances are added or removed, the Load Balancer updates its target group accordingly.
  • Ensures smooth traffic distribution at all times.

🧠 Example Scenario​

TimeExpected LoadASG ActionResult
DaytimeHigh trafficScale outAdd instances
NightLow trafficScale inRemove instances

⚡ Auto Scaling Strategies​

There are four main types of scaling strategies in AWS Auto Scaling Groups:

1️⃣ Manual Scaling​

  • User manually changes the desired capacity (e.g., from 1 → 2 instances).
  • Useful for testing or controlled environments.

2️⃣ Dynamic Scaling​

  • Automatically adjusts capacity based on CloudWatch alarms.

🔹 Simple or Step Scaling​

  • Triggered by specific CloudWatch alarms.

  • Example:

    • If CPU > 70% for 5 minutes → Add 2 instances.
    • If CPU < 30% for 10 minutes → Remove 1 instance.

  • "Step" scaling allows incremental scaling based on thresholds.

🔹 Target Tracking Scaling​

  • Simplified approach.

  • Define a target metric value, and ASG automatically adjusts to maintain it.

  • Example:

    • Target average CPU utilization = 40%.
    • ASG adds/removes instances automatically to maintain this value.

3️⃣ Scheduled Scaling​

  • Scaling actions are performed at specific times.
  • Ideal for predictable, time-based workloads.

⏰ Example:​

“Increase minimum capacity to 10 EC2 instances every Friday at 5 PM before a big event.”

4️⃣ Predictive Scaling (AI/ML Based)​

  • Uses Machine Learning to predict future traffic patterns.
  • Automatically adjusts capacity in advance of expected load changes.
  • Ideal for repetitive daily or weekly traffic spikes.

Example:​

If traffic spikes daily between 6–9 PM, ASG will automatically add instances at 5:50 PM.

🧭 Summary​

Scaling TypeTriggerAutomationExample Use Case
Manual ScalingUser input❌Testing or fixed capacity
Simple/Step ScalingCloudWatch alarm✅CPU spikes or drops
Target TrackingMetric target✅Maintain steady CPU utilization
Scheduled ScalingTime-based schedule✅Known traffic patterns
Predictive ScalingML-based prediction✅✅Regular, repeating patterns

🏁 Key Takeaways​

  • ASG ensures elasticity, cost savings, and fault tolerance.
  • Scaling strategies give flexibility — from manual control to ML-powered automation.
  • Load Balancer + ASG = reliable and scalable cloud application setup.
  • Remember: Elasticity = scaling in and out automatically based on demand.