# Deployment

In production, we recommend you use a process manager to spin up multiple workers, increase throughput and increase resiliency in case any of the workers fails.

# Running with Gunicorn

Gunicorn is a very popular option to manage multiple application processes in production. Luckily, uvicorn includes a worker class which means you can run your Bocadillo apps on Gunicorn with very little configuration (details: Uvicorn Deployment).

The following will start a Gunicorn server for your application:

gunicorn -w 4 -k uvicorn.workers.UvicornWorker -b localhost:8000 app:app

Let's break down this command:

1. gunicorn: this is the Gunicorn executable, which can be installed via pip install gunicorn
2. -w 4: Gunicorn will start and manage 4 application processes (a.k.a workers).
3. -k uvicorn.workers.UvicornWorker: here we pass Uvicorn's worker class
4. -b localhost:8000: specifies on which host and port the application should run.
5. app:app: identifies the Bocadillo application in the path.to.module:object format.

# Running with Docker

Bocadillo applications can easily be containerized to run in Docker containers. You can find an example Dockerfile in the docker-example repo.

# What about static files?

A typical answer to, "How should I service static files for my Gunicorn-served web app" is that you should use a reverse proxy such as Nginx. Even though this and other options such as using a CDN or object storage are valid approaches, they're difficult to get started with and require extra sysadmin work.

Bocadillo keeps it simple by using WhiteNoise, a library that allows your application to serve its own static files in a simple and performant enough manner, making it self-contained — and ready to be deployed on managed platforms.

In practice, this means that you won't need any extra steps to serve static files in production, unless you have very high performance requirements, in which case you should probably put your app behind a CDN.

# Deployment solutions

# Heroku

Bocadillo applications are very easy to deploy to Heroku. Check out the Heroku deployment guide to get started!