Immutable Infrastructure

Immutable infrastructure changes how you maintain production systems. You do not patch, reconfigure, or repair a running resource in place. You build a replacement, test it, deploy it, and retire the old resource.

“Immutable” describes the operating rule after deployment. It does not mean the system never changes. The system changes through replacement.

Why replacement changes operations

An in-place change starts with a running resource and alters its state. Over time, manual fixes, partial updates, and repeated configuration runs can make nominally identical resources behave differently. That difference is configuration drift.

Replacement starts from a version-controlled definition or image. The delivery pipeline builds the candidate and applies the same checks each time. Production receives the tested candidate rather than a sequence of repair actions.

This model gives you a clearer answer to a basic incident question: “What is running?” The deployed version points back to a build input and its test evidence. You can compare versions instead of reconstructing a server’s change history.