Medium · 12 steps · ~75 min

Design a food delivery app

Browse, order, track, all at dinner-time peak.

Restaurant lists are read constantly while orders are comparatively rare but must never be lost. Everything spikes at dinner.

Same problem as SwiggyZomatoDoorDash

How it works

• Browse: search and read restaurant menus.
• Order: record the order and notify the restaurant.

What you'll build

1. The API and the data model for restaurants, orders and deliveries
2. A system that answers 600 requests a second
3. Cached menus within a budget
4. A queue and workers that take orders through the dinner rush
5. A CDN for a nationwide promotion and a limiter against scrapers
6. A design that survives a cold cache, a lost worker and a lost primary

Steps

1. 1Design the API
2. 2Design the data model
3. 3Handle 600 requests/sec
4. 4Cache the menus for $900 a month
5. 5Dinner rush: queue the orders at 25,000 requests/sec
6. 6Friday night at 80,000 requests/sec for $2,200
7. 7How do people find 'pizza near me'?
8. 8Survive menu scrapers on $1,400
9. 9Survive a cold menu cache
10. 10Lose an order worker
11. 11Lose the primary database
12. 12Deep dive: tracking the courier live