Auto Scaling Basics

Learning outcomes

By the end of this lesson, you will be able to:

1. Explain what Amazon EC2 Auto Scaling does.
2. Define an Auto Scaling group.
3. Explain minimum, desired, and maximum capacity.
4. Describe how Auto Scaling works with health checks and load balancers.
5. Recognize target tracking as a straightforward scaling policy.

---

What Auto Scaling does

Auto Scaling helps your EC2 application automatically keep the right number of instances running. It ensures you have the correct amount of compute power to handle your application's load at any given time.

A simple memory rule:

• Load Balancer = Spreads the traffic.
• Auto Scaling = Changes how many instances exist.

---

1) What is an Auto Scaling group (ASG)?

An Auto Scaling group is a collection of EC2 instances that are managed as a single unit. Instead of managing individual servers, you tell AWS the "rules" for the group, and Auto Scaling handles the rest.

To create an ASG, you usually need:

• Launch Template: The "blueprint" (AMI, instance type, key pair) used to start new instances.
• Capacity Settings: The limits for how small or large the group can be.
• Scaling Policies: The logic that tells the group when to grow or shrink.

---

2) The Three Capacity Numbers

These three settings are the heart of how an Auto Scaling group behaves:

Minimum Capacity

The lowest number of instances the group will ever have. Even if there is no traffic, Auto Scaling will keep this many instances running for basic availability.

Desired Capacity

The number of instances the group tries to maintain right now. If an instance fails, Auto Scaling will automatically launch a new one to get back to this number.

Maximum Capacity

The absolute limit on how many instances the group can grow to. This protects you from unexpected cost spikes by capping your total instance count.

---

3) Self-Healing with Health Checks

Auto Scaling isn't just about traffic spikes; it's also about reliability. It continuously monitors the health of every instance in the group.

If an instance becomes "unhealthy" (e.g., the software crashes or the hardware fails), Auto Scaling will:

1. Terminate the unhealthy instance.
2. Launch a brand-new instance from your template.
3. Restore your desired capacity automatically.

---

4) Target Tracking: "Set it and Forget it"

How does the group know when to scale? The most straightforward method is a Target Tracking Scaling Policy.

You choose a metric (like average CPU usage) and a target value (like 50%). Auto Scaling then acts like a thermostat:

• CPU > 50%? It launches more instances to share the load.
• CPU < 50%? It terminates unneeded instances to save you money.

---

5) Better Together: ELB + ASG

When you pair a Load Balancer with an Auto Scaling Group, you get a professional, high-availability architecture:

1. Users connect to the Load Balancer.
2. Auto Scaling adds/removes instances based on traffic.
3. Load Balancer automatically starts sending traffic to new instances as soon as they are "healthy."

---

Micro-activity 1: Read the Group Settings

Micro-activity 2: Choose the Better Tool

---

Summary

Amazon EC2 Auto Scaling ensures your application has exactly the right amount of compute power. By using Auto Scaling Groups and Target Tracking, you can build applications that "self-heal" from failures and "self-scale" to handle any amount of traffic without manual intervention.

---

Knowledge Check

Next lesson

Lesson 4.7: Horizontal vs Vertical Scaling