System Design of Doordash: Geo-Hashing and WebSockets for Location Based Services

We go through a popular interview question: Design Doordash.

The system design of Doordash (similar to Swiggy and Zomato in India) involves matching food orders to riders. A rider has to be selected based on their location proximity to the restaurant (since the time taken to deliver a order from restaurant to customer wont change on rider).

For this, we shard geographical locations using hashes, known as Geo-Hashing. An area can be recursively broken down using a geo hash. Riders within a goehash region can be asked to pick an order on demand.

Tracking a delivery is done using server-side events or WebSockets. I suggested the idea of WebRTC for this, but it seems like overkill. Let me know your thoughts :D

Jordan's YouTube Channel: @jordanhasnolife5163

Location-based databases: https://youtu.be/OcUKFIjhKu0
System Design Website: https://interviewready.io

00:00 Intro
01:35 Functional Requirements
02:50 Capacity Estimations
06:45 API Endpoints
08:10 Data Sources
11:30 Onboarding a restaurant
12:20 GeoHashes
23:20 Driver Updates
27:00 Data Consistency
28:30 Consistent Hashing
36:40 Optimizing Deliveries
40:20 Delivery Tracking
44:44 WebRTC
46:18 Concluding thoughts

You can follow me at:
Github: https://github.com/coding-parrot/
Instagram: https://www.instagram.com/applepie404/
LinkedIn: https://www.linkedin.com/in/gaurav-sen-56b6a941/
Quora: https://www.quora.com/profile/Gaurav-Sen-6
Twitter: https://twitter.com/gkcs_

#SystemDesign #InterviewReady #Coding