TL;DR #

When your python code encounters an await keyword, it just means that it’s going to send the control back to the event loop to let other functions execute.

await lets the CPU know that a coroutine is waiting for an operation that isn’t CPU bound e.g. network requests, database writes, etc.

coroutines always executes sequentially and the event loop is what manages the execution order of coroutines (via asyncio.gather, asyncio.create_task)

await forces asynchronous operations to be completed in series - meaning they are executed as if they were synchronous functions. This is necessary if the result of the next operation depends on the result of the last one.

await is a checkpoint where it’s safe for asyncio to go to another coroutine

For example,

async def display_user():
    user_details = await get_user_details()
    profile_details = await get_profile_details(username=user_details.username)

    print(user_details)
    print(profile_details)

get_profile_details requires a username, therefore is dependent on the previous result. That’s why we added an await. But if they are both independent, it’s better to create a task and use asyncio.gather

async def dislpay_users_and_locations():
    task_users = asyncio.create_task(get_users())
    task_locations = asyncio.create_task(get_locations())


    users, locations = await aysncio.gather(task_users, task_locations)
    print(users)
    print(locations)

Another example

async def get_users():
  users = await session.execute('select * from users')
  return users

asyncio.run(get_users())

the await keyword here tells the asyncio or event loop: While we wait for the response, feel free to do other stuffs in the mean time

Resources: