Asynchronous Python API applications - Playground

Table of Content (ToC)

• Asynchronous Python API applications - Playground
• Table of Content (ToC)
• Overview
• Useage
• Installation
  • Python
  • The Uvicorn playground application

Table of contents generated with markdown-toc

Overview

That project aims at understanding, by practicing, how the asynchronous part of Python API applications work.

To the best of our knowledge, the only asynchronous framework to power Python applications is Uvicorn, which is also the underlying framework used by FastAPI, itself a nice framework to operate Python-based API applications.

The starting point of this project is a discussion on StackOverflow about the respective performances of asynchronous versus synchronous calls.

It is indeed not enough that FastAPI (and Uvicorn) are asynchronous frameworks, the developers have to explicitly make their code asynchronous, by using the await Python command (and specifying the API endpoint callback functions with the async prefix). But not all Python payloads may be made asynchronous-ready.

This project is an attempt to explore what asynchronous means for API endpoints functions.

Useage

• Launch the server
  • If Uvicorn and FastAPI are already installed in your pip environment (see the installation section below if needed):

$ uvicorn uvicorn-delay:app --port 8001 --reload
INFO:     Uvicorn running on http://127.0.0.1:8001 (Press CTRL+C to quit)
INFO:     Started reloader process [21835] using statreload
INFO:     Started server process [21870]
INFO:     Waiting for application startup.
INFO:     Application startup complete.

• With pipenv:

$ pipenv run uvicorn uvicorn-delay:app --port 8001 --reload
...
INFO:     Application startup complete.

• From another terminal window/tab, query the API with a client, for instance cURL:

$ curl localhost:8001/delay/0.0/0.0 && echo

{"delays":[0.0,0.0],"total_time_taken_ms":5702,"times_avarage_ms":42.69,"times":[146,70,47,42,44,40,31,30,30,69,71,40,40,40,41,40,41,39,40,40,40,40,39,40,40,49,40,40,40,39,40,39,40,39,39,39,39,39,39,44,58,58,43,41,39,38,38,39,39,38,40,39,39,41,39,39,39,51,54,70,43,40,39,39,39,39,38,38,39,39,39,38,38,39,39,38,39,55,63,43,39,38,39,39,39,39,42,41,40,38,39,39,38,38,38,39,39,38,38,38]}

• Or from a browser with the https://localhost:8001/docs URL:

Installation

Python

• If not already done so, install PyEnv. For instance
  • Prepare the environment (here, with the Bash as default Shell):

$ git clone https://github.com/pyenv/pyenv.git $HOME/.pyenv
$ cat >> ~/.profile << _EOF

# Python
eval "$(pyenv init --path)"

_EOF
$ cat >> ~/.bashrc << _EOF

# Python
export PATH="\${HOME}/.pyenv/bin:\${PATH}"
. ~/.profile
if command -v pyenv 1>/dev/null 2>&1
then
        eval "\$(pyenv init -)"
fi
if command -v pipenv 1>/dev/null 2>&1
then
        eval "\$(pipenv --completion)"
#       echo
fi

_EOF
$ . ~/.bashrc

• If PyEnv was already cloned a long time ago, think to update the repository from time to time:

$ pushd ~/.pyenv && git pull && popd

• Install a specific version of Python (e.g., as of August 2021, 3.9.6 is the latest stable version):

$ pyenv install 3.9.6 && pyenv global 3.9.6

• If not already done so, install pip and pipenv:

$ pip install -U pip
$ pip install -U pipenv

The Uvicorn playground application

• Clone this repository and change directory to it:

$ mkdir -p ~/dev/showcase && \
  git clone https://github.com/cloud-helpers/uvicorn-playground.git ~/dev/showcase/uvicorn-playground
$ cd ~/dev/showcase/uvicorn-playground

• If not already done so, install a few Python required modules
  • Either with pipenv (so as to insulate that project from anything else):

$ pipenv install

• Or with pip directly (which may mess with your Python installation):

$ pip install -r requirements.txt