System requirements

Functional:

• Able to generate a short URL given a long URL, a short URL could only represent one long URL
• Short URLs can be designed to expire after a specified duration
• 100k URLs, 500 times a day.

Non-Functional:

• High available
• Low latency of getting short URLs

Capacity estimation

• 5,000,000 accesses per day
• 10M QPS

API design

GET /short_id

POST /add_uri

{

long_url = long_url

}

Database design

Short_URL

id, long_url, expireation_timestamp

High-level design

API gateway for rate limiting

Backend server for generating tiny uri, persisted in databases cache and serving write traffic

Clean-up service for cleaning up the expired entries in the database

Request flows

• Write flow:
• the client sends a request, inserts into db with retry using a different key
• Update the cache
• returns
• Write flow:
• the client sends a request, server will find a hash,
• If found and not expired return 302 with long URL
• if not found and expired, query the db
• if found and not expired return 302
• else 404
• Maintenance flow: on db hook, send a queue task to schedule clean up, there will be a corn job per day to clean up expired entry.

Detailed component design

• Short URL generation design
• (a) hash
• (b) random string with db dedup
• (c) automatic incremental

Trade offs/Tech choices

Failure scenarios/bottlenecks

Try to discuss as many failure scenarios/bottlenecks as possible.

Future improvements

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