Stage 1 of 5 · Model
The location stream is the first bottleneck
Start with the data that arrives whether or not anyone requests a ride. There are 60,000 online drivers, and every available driver sends one ping every three seconds. Ride requests average 2,000 a minute and briefly peak at five times that rate.
System so far· 2 parts
Select a component to see what it is responsible for and which state it owns.
What you need to know
0 of 1 checks done
For a periodic stream, divide the number of active senders by the reporting interval. Do not size a system only for the user action that looks most important: background location updates can outnumber ride requests by orders of magnitude.
Work it out
60,000 drivers each send one location ping every 3 seconds. About how many pings per second reach the service?