For the past 9 months, I’ve been working with a client building a cloud brokerage system - one that abstracts away the provisioning of infrastructure to our client, and allows them to swap between AWS, Azure, etc, quite easily. We’re using a modern stack of devops tools such as Terraform, Ansible, Vault to create and maintain these environments, so we’re in a great position from a dev side. It raised some interesting questions from a functional test perspective, though - if our system’s function was to create and configure infrastructure, well, how do we test that all that?

Our solution was to create a separate functional test suite of infrastructure tests, alongside all our other testing frameworks. It worked pretty well, but we noticed in the meantime that this seems a pretty rare thing to do. Which is odd, because our infrastructure is code - why on earth aren’t we also testing it? It seems that potentially we’re developing a testing blind spot.

It’s reasonably generalist. There are some specifics in terms of our implementation, but I don’t want to get bogged down in the details - how you test your infrastructure most likely depends upon what your infrastructure is.