Katkat
Back

Design a food delivery system

CoinbaseSoftware Engineer · Technical Screen
MediumSoftware EngineerTechnical ScreenSystem Design

Last updated

This problem tests your ability to design a distributed, real-time logistics platform, focusing on high-concurrency order placement, real-time courier tracking, and transactional integrity during payment and dispatch.

Problem Statement

Design a food delivery system (similar to UberEats or DoorDash) that connects customers, restaurants, and couriers. The system must handle high-concurrency order placements, real-time location updates for couriers, and the matching of orders to available couriers.

Requirements

Functional

  • Customers can browse restaurants and place orders.
  • Restaurants can manage their menu and receive order notifications.
  • Couriers can accept delivery requests and update their real-time location.
  • The system must dispatch orders to the nearest available courier.

Non-Functional

  • High availability and scalability during peak hours (e.g., Friday evenings).
  • Low-latency real-time updates for courier location tracking.
  • Strong consistency for payment processing and order status transitions.

Your Task

Describe the high-level architecture, key services, and database schema. Provide details on how you would handle:

  1. The real-time courier tracking feature (e.g., how to handle frequent location updates).
  2. The dispatching algorithm (how to find a courier efficiently).
  3. Handling payment idempotency so a user is not charged twice for the same order.

Design Constraints

  • Assume 100,000 concurrent active orders and 50,000 active couriers.
  • Focus on the choice between SQL and NoSQL for different services (e.g., Order service vs. Tracking service).
  • Explain how you would ensure the system remains resilient if the notification service goes down.

Hints

Hint 1

Consider using a geospatial database or a quad-tree index to store courier locations to optimize the 'nearest courier' search.

Hint 2

Separate the read-heavy path of menu browsing from the write-heavy path of order placement and state transitions.

Hint 3

Use a message queue or event-driven architecture to decouple the order status update from downstream processes like restaurant notifications or customer push notifications.

Follow-up questions

  • How would you handle the 'Thundering Herd' problem if thousands of customers attempt to order from the same restaurant at the exact same moment?
  • If a courier's phone battery dies mid-delivery, how would you design the system to handle the state transition and re-assignment of the order?
  • How would you design a batching system to allow a single courier to pick up orders from multiple restaurants in a single trip?