Suppose we have a web server that lets us query some data or save it. This web server serves multiple requests at the same time, and each request is a short task that involves some computation. What is the usual way of achieving this? Well, the naive way would be to spawn a new thread for each request. But one can easily realize that this leads to an explosion in the number of threads in the application. Plus, the creation and deletion of threads are heavyweight activities; they slow down the entire application.

Next, you can use a thread pool so the same threads can be used over and over to avoid the overhead of creation and deletion of threads. However, if you want to serve thousands of requests at the same time, this will require a thread pool with thousands of threads. Thread scheduling in an operating system is complex and involves a lot of logic, including priority and so on. Operating systems do not expect threads to just run short bursts of computation...