Stream Limit Service

This approach uses Express middleware to validate when a user can watch a new stream.

Resources

Middleware are applied to the following endpoints:

	Endpoints	Description	Required	Middleware
GET	/api/play/:clipId	Play clipId content	Authorization header or token param on URL	addToPlaybackList isPlaylistLimit
POST	/api/user/logout	Logout user	Authorization header or token param on URL	removeFromPlaybackList

Authorization

• Authentication implementation is ignored and it works providing HTTP Authorization request header or token param:

curl -X GET \
  http://127.0.0.1:3000/api/play/CLIP_ID \
  -H 'Authorization: Bearer eyJlbWFpbCI6ImFubmFAbWFpbC5jb20ifQ=='

curl -X GET \
  'http://127.0.0.1:3000/api/play/7a7dc4ae-549d-481b-b7d8-46ef02981745?token=eyJlbWFpbCI6ImFubmFAbWFpbC5jb20ifQ=='

• Authorization schema type Bearer and credentials contains base64 encoding VALUE for email (acting as user ID)

Middleware

• addToPlaybackList This service will track every new request to watch a content into openConnections variable.
• isPlaylistLimit This service will return { status: 403, message: 'FORBIDDEN' } response when user exceeds more than 3 requests to play content.
• removeFromPlaybackList: When user logout a device, it will decrease reference to open connections.

Scaling

• We could make use of Node clustering to scale the application execution on multiple processor cores by creating worker processes to solve some issues such as HTTP request bottlenecks.
• Load balancer: we can run multiple instances of a Node application and use an NGINX server to proxy requests/connections.

Monitoring

• Detecting and preventing downtimes via log monitoring tools like Keymetrics, New Relic.
• Collect memory usage to find out memory leak issues

Running service and test

• Server on port 3000

npm install
npm run dev
npm test

Some assumptions

• For simplicity, this exercise is only faking one DB query response.