The Genius System Behind the Uber App’s Real-Time Map

Uber's real-time map system evolved from inefficient polling (80% of requests) to a push-based architecture called Ramen, built on gRPC. The system uses three components: Fireball (decides when to push), API Gateway (enriches data), and Ramen server (delivers updates). To handle millions of location updates, Uber developed H3, an open-source hexagonal spatial indexing system that reduces time complexity from O(N) to O(K²) + O(M) by dividing the world into hexagons instead of calculating distances to all drivers. Edge servers minimize latency by serving users from geographically close locations, while dead reckoning and Kalman filters smooth location updates during network instability.