Using Auto Scaling in AWS

Task 1 — Build the custom AMI from the CLI

Working from the Command Host. The Vocareum lab provides AWS credentials via an IAM role on the host itself, so aws configure only sets region and output format — no access keys.

• Confirmed the region using the EC2 Instance Metadata Service: curl http://169.254.169.254/latest/dynamic/instance-identity/document | grep region → us-west-2.
• Ran aws configure and accepted defaults for region (us-west-2) and output (json); access keys were left blank intentionally.
• Inspected UserData.txt with more to understand exactly what would be baked into the AMI before launching the instance.

#!/bin/bash
yum update -y --security
amazon-linux-extras install epel -y
yum -y install httpd php stress
systemctl enable httpd.service
systemctl start httpd
cd /var/www/html
wget http://aws-tc-largeobjects.s3.amazonaws.com/CUR-TF-100-TULABS-1/10-lab-autoscaling-linux/s3/ec2-stress.zip
unzip ec2-stress.zip

echo 'UserData has been successfully executed.' >> /home/ec2-user/result
find -wholename /root/.*history -wholename /home/*/.*history -exec rm -f {} \;
find / -name 'authorized_keys' -exec rm -f {} \;
rm -rf /var/lib/cloud/data/scripts/*

The last three lines matter: they wipe shell history and any stray authorized_keys before the AMI is captured, so the image is not contaminated with the original instance's credentials or identity.

# Launch the WebServer instance with UserData, tag, and public IP
aws ec2 run-instances \
  --key-name vockey --instance-type t3.micro --image-id ami-03bc3367c1f063162 \
  --user-data file:///home/ec2-user/UserData.txt \
  --security-group-ids sg-01fe8a4a5dcf71ff4 \
  --subnet-id subnet-077230f8b0253fec5 \
  --associate-public-ip-address \
  --tag-specifications 'ResourceType=instance,Tags=[{Key=Name,Value=WebServer}]' \
  --output text --query 'Instances[*].InstanceId'
# → i-0cdb511e7caffe386

# Resolve its public DNS once the instance reports the network interface
aws ec2 describe-instances --instance-id i-0cdb511e7caffe386 \
  --query 'Reservations[0].Instances[0].NetworkInterfaces[0].Association.PublicDnsName'
# → "ec2-54-185-40-112.us-west-2.compute.amazonaws.com"

# Capture the running instance as a custom AMI (reboots by default for FS integrity)
aws ec2 create-image --name WebServerAMI --instance-id i-0cdb511e7caffe386
# → { "ImageId": "ami-0c9afcf8dc58a824a" }

Task 2 — Application Load Balancer in two AZs

Created from the EC2 Console under Load Balancers. Picked Application Load Balancer because the workload is HTTP and benefits from layer-7 routing and health checks per path.

Task 3 — Launch Template + Auto Scaling Group

Two console resources here. The Launch Template is the immutable blueprint for new instances; the ASG is the controller that uses it to keep the right number of instances alive.

Task 4 — Drive load and observe scale-out

With targets reporting healthy, the ALB's DNS name serves the bundled PHP page that exposes two buttons — Start Stress and Stop Stress — wired to the stress command on the instance that handled the request.

Start Stress launches the stress binary on whichever instance answered the request, pushing its CPU to ~100%. CloudWatch averages CPU across the ASG; once the average crosses 50% for the cooldown window, the target tracking policy adds new instances up to max=4. Stopping the stress brings the average back down and the ASG scales back to desired=2.