EC2 Autoscaling - Cloud Pre-Kubernetes

Elasticity may be the most critical feature of any public cloud. EC2 Autoscaling is the OG of autoscaling.

You have a batch process that needs variable amounts of compute capacity each time it’s run. So, you leverage Lambda, EKS, or another container-based orchestration layer on AWS, right? What if that batch process has to run in a virtual machine instance? Well, there’s EC2 Autoscaling. EC2 Autoscaling is elasticity before we had Kubernetes or Lambda. The concept is a fundament of cloud elasticity.

In a nutshell, you set an OS threshold or application metric, and EC2 Autoscaling orchestrates your AWS infrastructure to meet that threshold. You always want 3 instances of your web server running, and there’s a known bug that creates a memory leak? You can have autoscaling take the “failed” instance out of your group and start up a new EC2 instance. I’d much rather you bother the developer to fix the leak, but let’s be honest here.