One of the common things I’ve seen in automation implemented by infrastructure teams is a lack of rigor around testing. That is to say, code that tests the task that is being automated is actually successful. A script could execute to its end without errors, but that doesn’t necessarily mean it actually did what it was supposed to.
This leads into a concern I’ve seen raised by stakeholders, about visibility of what is happening in an automation pipeline. One of the key stakeholders in this sort of security is the IT Security team, who often want visibility of certain outputs (like virtual machines) to determine if those are secure. In a couple of environments I’ve raised the idea of performing automated vulnerability scans on newly provisioned assets, as a way of ensuring what is delivered is at an acceptable level. By automating this task, we place no extra burden on those involved and ensure consistency.
