Video narration
Imagine 10,000 people open your SaaS at the same moment. First stop is the load balancer—like a traffic cop—spreading requests across many application servers so no single server melts. Those app servers handle fast work immediately, and they offload slow tasks—like sending emails or generating reports—into a job queue. Workers pull jobs from that queue and process them in the background, smoothing out spikes. When traffic rises, autoscaling can add more app servers and more workers, so your “front of house” keeps up. But every request still leans on the database for reads and writes. The bottleneck is simply the narrowest part of the system. If the database can’t keep up, it becomes that narrow point: queries pile up, app servers start waiting, pages feel slow or time out, and adding more app servers can actually increase the pressure. At that point you must relieve the database—scaling apps isn’t enough.