The first time you stand up an ALB and hook two EC2 instances behind it, it feels like magic. You fire requests at a single URL and half of them land on server A, the other half on server B. When CPU spikes, the ASG spins up more instances on its own; when one dies, it launches a replacement. Behind that magic, a health-check algorithm, the target group's internal state machine, and the ASG's reconciliation loop are all running with precision.
Today we look at exactly which OSI layer ALB and NLB make their routing decisions on, how an ASG turns a fresh instance into "InService", and how Blue/Green deployment is implemented with ALB target group weighted routing