How would you design a system that scales to millions of users on AWS?

Step 1: Use Cases, Constraints, Assumptions, Back-of-Envelope

Use cases

We'll scope to handle only the core flows that expose the bottlenecks; everything else is deferred.

• User signs up / logs in — auth on stateless API, session in Redis, email verify async.
• User reads feed/profile — cache-first, CDN for assets, read replicas for DB.
• User writes (post/like) — write to master, async fan-out, eventual read visibility.
• Service scales — milestone Users+ → Users+++++, each step adds one lever, measured.

Out of scope

• Multi-region active-active (defer), complex ML ranking, payments (separate PCI subsystem).

State assumptions

• Start 10k DAU → 10M DAU iterative; final 10M DAU, 5% write, 95% read.
• Avg request 10 KB in / 20 KB out, 10 requests/user/day → 100M requests/day.
• Write QPS: 10M*0.5/86k≈58 avg, 500 peak; Read QPS: ~1,100 avg, 10k peak.
• DB