Uber Architecture

Real-time matching of riders and drivers across the globe.

On the roadmap:Software Architect · step 5: Learn from the giants

Overview

Uber's platform handles geo-spatial dispatch, pricing and payments in real time. It evolved from a monolith into thousands of microservices, with custom datastores built on MySQL.

Architecture highlights

Geo-sharded dispatch for low-latency matching
Schemaless: append-only datastore on sharded MySQL
Kafka as the backbone for event streaming
Domain-oriented microservice architecture

Architecture diagram

Key components

Dispatch service
Matches riders with nearby drivers
Location pipeline
Ingests driver GPS pings continuously
Schemaless / Docstore
Sharded durable storage
Kafka
Event bus for trips, payments and analytics
Pricing service
Computes surge from supply and demand

Challenges and trade-offs

Geo-spatial indexing
Finding nearby drivers requires efficient spatial indexes such as H3.
Write-heavy location stream
Millions of location updates per second need partitioned ingestion.
Consistency of trips
A trip must never be matched twice; state transitions need strict ordering.

Study: Uber

Learn how Uber is built from the people who explain it. Watch first for the picture, then read for the detail.

0 of 5 watched

16 min

Start here

Basic System Design for Uber or Lyft | System Design Interview Prep

Interview Pen · 899K views

Then

Cookie preferences