Architecting for Scale: A Founder’s Field Guide

Scaling isn’t magic; it’s discipline. Here’s a short list of decisions that save you when traffic shows up.

Pick Boring Tech (On Purpose)

• Postgres + Redis + REST/GraphQL — fewer moving parts, fewer surprises.
• Background jobs for anything slower than ~200ms to users.

Schemas and Limits

• Declare limits early (max upload size, rows/page, retries).
• Add created_at indexes you’ll actually query on.
• Prefer explicit states over booleans; it helps during incidents.

Observability You’ll Use

• Structured logs with request IDs; log why, not just what.
• One latency budget; one error budget; page on budget breaches.

You don’t need “planet scale”. You need a playbook you’ll remember at 2am.