Short polling
- steps
- client sends a request
- server responds immediately with a handle (usually a unique identifier that corresponds to the request)
- the server do whatever - queue, persist through disk, put it in memory & execute etc. The request is not executd immediately
- server continues to process the request
- client uses that handle to check for status
- Multiple “short” request/response as polls
- so poll itself is a response model, but the entire system is asynchronous.
- The response can be done anytime, but it’s sent back to client only when client polls
- Pros
- Simple
- good for long running requests
- client can disconnect safely
- Cons
- Too chatty… imagine you scaled up & u have 1000s of ppl. most of them will be useless where the backend responds with
false lol
- network bandwidth
- this is precious in the backend.. since if you put ur architecture in cloud that’s how you get billed
- basically wasted backend resources
- the backend checking if the request is finished uses finite resources, and it could have been used for doing actual requests