python programming
Asyncio is specifically designed for managing states of IO-bound programs (you
can think of it as a successor to Twisted).
Link: https://www.reddit.com/r/Python/comments/8kzxm5/gil_and_asyncio/
The problem asyncio solves is that it allows you to use non-blocking IO
primitives with an (arguably) nicer programming interface which abstracts away
the need to loop through non-blocking sockets with select() and instead builds
it into an event based system that the program can construct itself around,
where these select() polls ultimately result in events being fired and
different "async" tasks in the Python program waking up. However, as you've
noted with asyncio, the event queue runs in only a single thread.
Asyncio doesn't block on socket, so it is only useful if your workload is
dominated by network access time. Ex: servers, clients making a lot of request,
etc.
Multiprocessing will make use of several cores, but process communication is expensive. You use it when your workload is mainly CPU intensive with a work you can easily split with little communication between the workers.
Threads never block, but share one CPU, and are useful when:
- file access dominated workload
- lot of local socket access (db access)
- avoiding blocking a GUI
- like asyncio, but you use some blocking libs