Infrastructure as Code, 2nd Edition

Chapter 19. Delivering Infrastructure Code

The software delivery life cycle is a prominent concept in our industry. Infrastructure delivery often follows a different type process in an Iron Age context, where it can arguably be less rigorous. It’s common for a change to be made to production infrastructure without being tested first; for example, hardware changes.

But using code to define infrastructure creates the opportunity to manage changes with a more comprehensive process. A change to a manually built system might seem silly to replicate in a development environment, such as changing the amount of RAM in a server. But when the change is implemented in code you can easily roll it across a path to production using a pipeline (see “Infrastructure Delivery Pipelines”). Doing this would not only catch a problem with the change itself, which might seem trivial (adding more RAM has a pretty small risk of breaking something), but would also catch any issues with the process of applying the change. It also guarantees that all environments across the path to production are consistently configured.

Delivering Infrastructure Code

The pipeline metaphor describes how a change to infrastructure code progresses from the person making the change to production instances. The activities required for this delivery process influence how you organize your codebase.

A pipeline for delivering versions of code has multiple types of activities, including build, promote, apply, and validate. Any ...