Design: Uber

// Driver updates location every 5 seconds
redis.geoadd("drivers:available", longitude, latitude, driverId);

// Find drivers within 2km of rider
List<GeoRadiusResponse> nearbyDrivers = redis.georadius(
    "drivers:available",
    riderLong, riderLat,
    2.0, GeoUnit.KM,
    GeoRadiusParam.geoRadiusParam()
        .withDist()  // include distance
        .count(10)   // max 10 drivers
        .sortAscending() // nearest first
);

public DriverMatch findDriver(RideRequest request) {
    List<Driver> nearby = locationService.getNearbyDrivers(
        request.origin, 2.0 /* km */, 10 /* max */);

    for (Driver driver : nearby) { // sorted by distance
        // Try to lock driver (prevent double-booking)
        if (redis.setnx("driver:locked:" + driver.id, request.id) == 1) {
            redis.expire("driver:locked:" + driver.id, 30); // 30s lock
            notifyDriver(driver, request);
            return new DriverMatch(driver, request);
        }
    }
    // No drivers available — return to queue for retry
    return null;
}

Driver App → WebSocket → Location Service → Redis GEO
                              ↓ also
                         Kafka (for trip tracking, analytics)
                              ↓
                         Cassandra (trip history)

// Surge multiplier based on supply/demand
public double getSurgeMultiplier(GeoCell cell) {
    double demand = getActiveRequests(cell);  // riders requesting
    double supply = getAvailableDrivers(cell); // idle drivers
    double ratio  = demand / supply;

    if (ratio < 1.2) return 1.0;  // normal pricing
    if (ratio < 1.5) return 1.5;
    if (ratio < 2.0) return 2.0;
    return Math.min(ratio, 3.0);  // cap at 3x
}

Geohash divides Earth into rectangular cells at various precisions:
Length 4: ~39km × 20km
Length 6: ~1.2km × 0.6km  ← good for driver density
Length 8: ~38m × 19m     ← precise driver location

Surge pricing uses length-5 cells (~5km) to aggregate supply/demand

trips: trip_id, rider_id, driver_id, origin, destination, status, fare, surge_multiplier, created_at
drivers: driver_id, name, license, rating, car_info, is_available
location_history: driver_id, lat, lng, timestamp (Cassandra time-series)