Every user is sent to Server A, and a big crowd arrives.
The line of waiting requests keeps growing.
Replies get slower and slower.
Server A runs out of memory and crashes. The site is down.
New users get errors, while two spare servers sit idle.
One server has a limit, and a crowd can break it.
A load balancer shares requests across many servers.
It checks health and stops sending traffic to dead servers.
Reach for it when
When one server cannot carry the traffic, or its crash would take the site down. Not for a one-server hobby app.
The catch
Everything flows through it, so if the only balancer dies, three healthy servers become unreachable. Run two.
Say it in an interview“I would put a load balancer in front of stateless servers. It spreads traffic, health-checks them, and lets us add servers or lose one without downtime.”