Asynchronous Python

The central focus of Asyncio is to perform multiple concurrent tasks that involve waiting period at the same time. The key insight for asynchronous programming is that while you wait for this task to finish, other tasks can execute.

Threading vs Asynchronous

Due to the GIL(Global Interpreter Lock) in Python, only one thread is executed at any time, there is no multi-core parallel processing for both multi-threading and asynchronous programming. However, we can still scale up the I/O blocking tasks with the two programming styles.

For multi-threading programming, each thread is instantiated for one task. While one thread enters into the waiting phase, it is released and another thread starts to run. While for asynchronous programming, only one thread runs all the time, every time it starts to wait, it switch to the next task. In the latter case, you have one thread that loops over all different tasks. So simply an event loop executes a collection of tasks.

The synchronization of different threads is one of the main difficulties in multi-threading programming. This is where the collision happens among different threads. With asynchronous programming, we can work around this by switching the tasks explicitly with Python coroutine. This is much safer.

Another advantage of asynchronous programming over multi-threading is the memory usage. Threads require a lot of memory, about 8MB per executing thread. This can consume a lot of memory when you run a thread per unique activity. While threads are costly to start, the cost of starting a generator coroutine is just a function call.