πŸ’¬
βœ‰οΈ
πŸ“Έ
▢️

Loading…

Design Uber: location, matching, and trip service

UberGuided Mode

Step 1 β€” Naive: Client β†’ API β†’ DB

Request Flow

1

Driver writes GPS

Driver app sends HTTP POST with lat/lng to the API Server every second. The API writes to the Database. 100K drivers Γ— 1/s = 100K writes/second β€” already at the DB limit.

2

Rider requests ride

Rider opens app, taps 'Request Ride'. The API Server runs: SELECT * FROM drivers ORDER BY ST_Distance(pos, $rider_pos) LIMIT 5 β€” a full table scan on 100K rows, every request.

3

Why it breaks

At 300 requests/second each doing a full distance scan on 100,000 driver rows, the Database is overwhelmed in minutes. Two fundamental problems: too many writes AND too-slow spatial reads.

Why We Need This

The DB is doing two conflicting jobs: absorbing 100K location writes/second AND answering nearest-driver spatial queries. No general-purpose relational DB handles both at this scale.

Key Insight

The moment you have moving entities and spatial queries, you need dedicated spatial indexing β€” not a general-purpose database trying to do everything.

Step 1 of 8