AWS ELB and Auto Scaling: high availability and elasticity

The Tabaski sales are coming, and Baobab Market's shop runs on a single server: a single point of failure. Awa asks you to make sure the shop stays open if a server goes down, adds servers when the crowd arrives, and removes them when the crowd leaves. This fourth lab of the preparation path for the AWS Certified Cloud Practitioner certification starts with the words the exam uses most to talk about the benefits of the cloud: high availability, fault tolerance, horizontal and vertical scaling, elasticity. An architecture diagram shows what you are going to build: a load balancer present in two Availability Zones, in front of an Auto Scaling group driven by CloudWatch. You read the provided launch template, then create a target group and its health check, an Application Load Balancer and its listener, and an Auto Scaling group with a target tracking policy, after a first denial triggered on purpose. The shop responds at a single address, served in turn by two instances. You kill one: customers see nothing, and Auto Scaling replaces it on its own. You then go up to three instances, and watch the policy scale back to two, triggered by a CloudWatch alarm it created itself. Everything happens in the real AWS console, with no Elastic IP address. The guide is illustrated with screenshots of the console and an architecture diagram, compares ALB, NLB and Gateway Load Balancer, explains the ways to scale and what a load balancer and public IPv4 addresses cost, and ends with a ten-question practice exam.