Python Async/Await Patterns

Overview

Use asynchronous programming for I/O-bound work: network requests, database queries, file operations, and inter-process communication. These operations spend most of their time waiting, and asyncio allows other work to proceed during that wait.

Use synchronous (or multiprocessing-based) code for CPU-bound work: image processing, cryptographic hashing, data compression, and heavy computation. The Python GIL prevents true parallel execution of Python bytecode in threads, so CPU-bound work does not benefit from asyncio.

A useful heuristic: if the bottleneck is waiting, go async. If the bottleneck is computing, use multiprocessing or a task queue.

Core Concepts

Three building blocks underpin all async Python code:

• Coroutine -- a function declared with async def. Calling it returns a coroutine object that must be awaited or scheduled.
• Event loop -- the scheduler that runs coroutines, handles I/O callbacks, and manages timers. asyncio.run() creates and manages the loop.
• Awaitable -- anything that can appear after await: coroutines, Task objects, and Future objects.