What Ansible is & the push model

Agentless automation over SSH.

Ansible is an automation tool for configuring servers, deploying software, and orchestrating changes across a fleet — and its defining trait is that it is agentless. There is no daemon to install on the machines you manage; a control node connects to them over ordinary SSH (or WinRM for Windows), pushes the tasks, runs them using the target’s Python, and disconnects. If a host has SSH and Python, Ansible can manage it.

That push model is the opposite of agent-based tools like Chef or Puppet, where each node runs an agent that pulls its config from a server. Agentless means almost nothing to bootstrap and a tiny attack surface on the managed hosts — but it also means the control node holds the SSH keys and privilege to change every machine, which becomes the security story. Ansible describes desired state in YAML, so the same playbook run twice converges rather than repeating work.

What you write

Two artifacts define an Ansible setup: an inventory (which hosts exist, grouped by role) and playbooks (what to do to them). Everything is YAML and declarative — you describe the state you want (“nginx installed and running,” “this config file present”) rather than the shell commands to get there, and Ansible’s modules make it so, reporting whether anything actually changed.