Rust adventures - Async - Part 1
A while ago I realized that I was a visual learner which can be frustrating at times since some concepts might take me a bit longer to fully understand until I create the proper mental image(s) for it (or somebody else does it for me).
When I started wrapping my head around Async programming in Rust I felt like I was missing some of those images. What follows is my attempt visualize the concepts around async programing.
Traditional threaded applications
When creating multi-threading applications, if you wanted to execute multiple tasks at the same time you'll need the same number of threads as tasks. So you'll have a 1-1 mapping between tasks and threads. Let's imagine for example, we wanted to execute 3 tasks:
It might looks like this at the Operating System (OS) level1:
During the execution of these threads, any of them can become blocked ( unable to continue to do work ) or they could just voluntarily yield ( programmer knew the thread will have nothing to do for some time under certain condition ), at that point the OS will have to save the state of the running thread (so that it can resume running at a later point) and select another to start or continue running. This is called thread-context switch.
For a variety of reasons2, this context switching has some amount of overhead that penalizes performance. Additionally, the approach doesn't scale well since the more tasks we want to run at the same time the more threads we'll need (eventually running out of resources).
With Asynchronous code instead of using one thread per tasks, we can run multiple tasks concurrently on the same OS thread.
Because multiple tasks are running on the same OS thread, the issue of thread-context switching is greatly improved (less threads to switch around). Furthermore, since we're reusing the same OS threads for multiple tasks, we end-up using significantly less resources.
For both of those reasons our Async counterpart has the potential of being much faster:
This code is meant to act cooperatively in a single thread.
In the next part we'll be exploring the
Future trait and the surrounding Async programming ecosystem further.