4 lessons
Statelessness and why it matters
The one design property that decides whether you can ever run more than one copy of your app server.
Load balancers
The traffic cop in front of your fleet of app servers, how it picks who gets each request, and how it knows a server is unhealthy.
Auto-scaling and when to actually add servers
How systems add and remove capacity automatically based on real load, and the concrete metrics and thresholds that trigger it.
Horizontal scaling in practice
How stateless servers, load balancers and auto-scaling fit together into one working fleet, with containers as the practical delivery mechanism.