Design Ola's Driver Location Tracking System at Scale

Problem

Design the backend system that tracks the real-time location of 2 million Ola drivers across India and serves this data to the matching and dispatch systems.

Requirements

Functional:

  • Drivers send GPS location every 5 seconds
  • Nearest available driver lookup by location + vehicle type
  • Historical route replay for dispute resolution
  • Driver heatmap for demand forecasting

Non-functional:

  • 2M active drivers at peak
  • 400K location updates/second
  • Nearest driver query < 100ms
  • Location history retained for 90 days

Discussion points

  1. Location ingestion: UDP vs WebSocket vs HTTP
  2. Geospatial indexing: Redis GEO commands, geohash, S2 cells
  3. Write-heavy storage: time-series DB (InfluxDB) vs append-only Cassandra
  4. Serving layer: in-memory grid for hot queries
  5. Historical storage tiering: hot (7 days) → warm (90 days) → cold
added …
LeaderboardSalaryAccount