Scaling One App from 1 User to 1 Million (and Beyond)
One Indian D2C store on NestJS + Postgres, taken stage by stage from a single VM to multi-region. Each stage gives the metric that forces the change, the change itself, what it costs or breaks, and a diagram.
About 85 minutes · 16 exam questions · 10 interview questions · 2 code labs
What you will learn
- The running example and the pizza-shop map of the whole journey
- Stages 0–1 — One box, then split the database off (1 to ~10K users)
- Stage 2 — Vertical scaling and where it stops (10K to ~50K users)
- Stage 3 — Stateless app tier behind a load balancer (50K to ~200K users)
- Stage 4 — Caching the hot reads (100K to ~500K users)
- Stage 5 — Read replicas for the reads you cannot cache (300K to ~1M users)
- Stage 6 — CDN and object storage for images and static assets (200K+ users)
- Stage 7 — Async work via queues: take side effects out of checkout (200K+ users)
- Stage 8 — Offload search from Postgres (500K+ users)
- Stage 9 — Partition first, shard much later (1M users and beyond)
- Stage 10 — Multi-AZ: survive losing a data centre
- Stage 11 — Multi-region: only with a reason, active-passive first
- The observability and cost ledger across every stage
Next: Scalability, Availability, Reliability, Latency & Throughput