Docker And The Unix Philosophy

Everyone and their dog has heard of Docker and containerization by now. The technology does present some amazing new opportunities, but for me one of the most striking things is that it allows us to get back to the Unix philosophy - do one thing, and do it well.

A well-containerized application has the minimum content necessary to function. To build a full application, we compose a number of containers together. We also want to use the exact same container in all our environments, so that the code that we develop is the same code that we test, and is the same code that we deploy. Taken together, this encourages us to write smaller, focused applications built with composition and configuration in mind. This is the same approach taken so successfully in the early days of Unix with pipelines (though a composition of containers is clearly different to a one-way pipeline). That also happens to be the crux of the recent trend towards microservices.

Containers are no silver bullet, however. Breaking an existing application's composition and deployment into a collection of containers requires that you think carefully about how you deploy and configure your applications. For example, instead of building a VM and provisioning it with Ansible to make it ready to execute your code, you might need to think about how to pass an app's configuration to it at runtime with some combination of environment variables and volume-mapped configuration files, or perhaps you need some scripting so that the container can bootstrap itself if necessary. For example, the official MariaDB container contains an entrypoint script that will create an installation, database, and user when first run.

If you haven't tried Docker yet, see their 10-minute interactive tutorial.

This site uses cookies. Continue to use the site as normal if you are happy with this, or read more about cookies and how to manage them.