Configuration: Making a Server Ready

Provisioning gave Sam a server. But if he logged into it right now, he'd find it almost bare — a running machine with none of the things Pageturn needs to actually run. Creating the server was only half the job.

The other half is configuration management: installing the software, applying the settings, and putting the right files in place so the server is genuinely ready to run Pageturn. Like provisioning, it's done from a file rather than by hand — the same steps applied to every server, identically. Ansible is the best-known tool for this half.

Empty serverprovisioned, bare

→

Configurationinstall & set up

→

Ready serverruns the app

The empty server problem

A freshly provisioned server is a clean room with the lights on and nothing else. The machine runs, but Pageturn needs particular things present to work: the runtime that executes its code, a few supporting tools, specific settings, certain files in certain places. None of that arrives by default.

So between "the server exists" and "the server runs Pageturn" sits a list of setup work. Someone, or something, has to do that work on the bare machine — and doing it badly is its own source of the drift and inconsistency we saw two topics ago.

What configuration management does

Configuration management is automating that setup: capturing the install-and-configure steps so a tool performs them on the server instead of a person doing it by hand. Point the tool at the empty machine, and it installs the runtime, lays down the settings, and puts the files where they belong — turning a bare server into a ready one.

Think of a newly built house. Provisioning is the construction crew handing you the finished shell — walls, roof, a working front door. But you can't live in it yet: the plumbing has to be connected, the appliances installed, the furniture placed. Configuration is all of that fit-out work, the part that makes an empty structure into a place that actually functions.

Described in a file, applied to every server

The setup lives in a file, just as provisioning did. That's what makes it reliable: the same described steps get applied to every server, identically, whether it's one machine or fifty. This is the automation idea from Chapter 2 pointed at server setup — write the steps down once, let a machine repeat them exactly, and stop trusting a tired human to remember step fourteen.

It also means setup isn't a one-time event. The tool can be run again to re-apply the file and pull a server back in line if something on it has drifted, keeping every machine matching the same written description over time. Consistency isn't set once and hoped for; it's re-asserted from the file.

Provision versus configure

Now the two halves of the chapter sit side by side. Provisioning creates the infrastructure — it makes the server exist. Configuration makes it ready — it installs and sets up what runs on that server. One brings the machine into being; the other furnishes it for work.

Real teams often use both together: a provisioning tool like Terraform to create the servers, then a configuration tool like Ansible to make them ready. Keeping the two ideas distinct is the thing to carry forward — "create the machine" and "make it ready" are different jobs, done by different tools, and together they're what people mean by infrastructure as code.