System requirements

Functional:

• Users should be able to map long URL into short URL
• Users should be able to fetch long URL from short URL
• URL should be persistent for a certain period of time

Non-Functional:

• Availability
• Performance
• Scalability

Capacity estimation

• Around 5000 RPS for total of roughly 200 million URLs

API design

POST /longurl returns shorturl

GET /shorturl returns longurl

Database design

• url (id, short, long, createdAt)
• analytics(id, url_id (foreign key), other analytics )

High-level design

• Scalable, large blob database for storing urls since its not easily modified, high read low write
• Back end with # replicas based on RPS performance of each replica pointed to one database

Request flows

• Client requests long url
• Backend fetches long URL based on shortURL
• returns long URL

• Client posts long url
• Backend checks if long URL in db
• if so return short URL, else create and return new short URL

Detailed component design

• Some sort of hash algorithm to make sure the shortened URL is sufficiently randomized

Trade offs/Tech choices

• Trade off database performance for scalability
• trade off write/update capabilities in database

Failure scenarios/bottlenecks

• Backend RPS performance

Future improvements

What are some future improvements you would make? How would you mitigate the failure scenario(s) you described above?