Uber

• Use cases

• Functional requirements

• Non-functional requirements

  • High performance

  • Scalable

• Real world

• References

• Appendix - Past chart

  • TODO

Use cases

• Real time cab booking service like Uber, Lyft, Bolt.

• Real time hotel / restaurant search like Yelp.

• Target shoppers nearing a certain store for a marketing promotion.

• Hyper-local delivery system — dispatch delivery agents for restaurant order like Uber Eats.

Functional requirements

• Given a location by a client or user, our service should figure out certain number of nearby locations.

  • NN: Given a location, find the nearest k points

  • Range query: Retrieve all points within a specific spatial range.

• Update location

Non-functional requirements

High performance

• The location should be accurate so that the delay in update should response ≤ 200 milliseconds.

• We need to support 50000 read queries per second & 10000 write queries per second.

Scalable

• Our service is used world wide.

• As users grow in our system, the system should scale linearly without adding much burden.

Real world

• Unique GeoId in Twitter
• TODO: Uber Marketplace: Location Serving & Storage in the Uber Marketplace

References

• R tree / KD - tree
• Geohash vs S2 vs H3
• Redis data model based basic algorithm for finding available drivers
• 四火唠嗑
• Indepth system design discussion

Appendix

• A great percentage of the below post comes from this blog https://kousiknath.medium.com/system-design-design-a-geo-spatial-index-for-real-time-location-search-10968fe62b9c

Past chart

Schema design

Report design

Storage design

TODO

• Uber architecture and system design: https://medium.com/nerd-for-tech/uber-architecture-and-system-design-e8ac26690dfc