Scalability Design

Scaling problems are rarely solved by adding more of what you already have — they're solved by removing whatever forces every request through the same place. A stateless web tier scales by adding boxes; a single Postgres primary handling every write doesn't, no matter how many app servers point at it. The architecture decides which resource has to grow linearly with load and which one hits a wall first.

Design decides the ceiling long before anyone hits it. The question worth asking early is not "can we add more servers" but "what breaks first, and does adding more servers even help it."

Scale the thing that's actually the bottleneck — everything else is decoration.

1. Make the request path stateless wherever possible