Distributed Systems with Node.js

Chapter 6. Deployments

A deployment, in the simplest sense, is the movement of code from one location to another. With some platforms this is as simple as copying a bunch of files. For example, plenty of applications can be deployed by copying raw source code files like PHP, Python, and Perl scripts, and subsequent HTTP requests to a web server execute the updated files automatically. Static sites are typically deployed in the same manner. More complicated applications that run persistently require an additional step to stop and start a process. Examples of this include shipping Node.js source files, a compiled Go binary, or a Python script.¹

Modern applications should make themselves consumable by listening on a port (see https://12factor.net/port-binding for details). This is true whether an application is written in a platform that is traditionally invoked by a web server (like PHP, where you might include Apache and PHP inside of a Docker container) or if the application is written in Node.js (where the process listens for requests directly, hopefully with an external reverse proxy still involved). Sure, Node.js processes can be restarted when a source code file has been changed. Packages like nodemon and forever provide such functionality for making local development easier.²

In practice, a deployment is a much more formal process than “just copying some files.” The deployment process is usually made up of many stages, with the copying of application code being one of the ...