Skip to content

Latest commit

 

History

History
38 lines (30 loc) · 2.21 KB

design-question.md

File metadata and controls

38 lines (30 loc) · 2.21 KB

Designing Twitter

How would you design Twitter? How would you identify challenges in the product? How would you scale out the product for 1,000s of users?

  • Focus is usually on the Data Models
  • DO NOT mention specific technologies
  • Use the whiteboard for visuals

  1. Figure out area to focus on. Ask interview: "Hey, this product Twitter has a many different features involved in it. Is there a particular area you would like me to focus on?"

  2. Without specific direction, focus on the Data Model.

"How you would build out the database to reflect all the information in a given product?"

  1. Discuss why you would use a types of technologies and not just listing off specifics technologies.

"I would decide to use a javacript framework on the front but it would have to fulfill these requirements... (work well on moblie, easy render out a list of tweets, etc.). I would design a backend that was primarily as REST API. This would allow us to have centralize backend that is supporting the moblie apps and web framework."

  1. Above all, voice your thoughs through out the response/process.

Strategy

  1. Identify Two Core Features in the Product

This is where you can draw out the diagram of the core user interface (Tweet and Feed).

  1. Possible Implementation

This would be everything from web interface to sending/handling a request on the backend. Think about some database solutions (fields, data types, etc.).

  1. Identify and Address Difficulties

Discuss challeges around your core two features. What you need to figure out to make feature work.

  • Tweeting - What does a tweet look like in the DB?
  • Tweeting - How to make the #topic and @mention system?
  • Tweeting - How to implement retweets?
  • Feed - How to show interesting tweets at top of feed? Track popularity or views?
  1. Solution for Scaling

Discuss general stategies of scaling

  • Caching Doing expensive work one time (Server -> Database) storing the result to be resused next time its queued (Server -> Cache/Memory Store).
  • Server Architecture Create a load balancer that users make request to and that machine randomly assigns request to one of many other machines. Example of Horizontal Scaling