I don't find the Docker API particularly intuitive. For my use-cases so far I've found it much simpler to spin up a VM if I want to duplicate a particular environment.
I think where Docker tends to shine is when you are operating at scale. For larger players using VMs to recreate a computing environment is expensive and eats into your bottom line. In such situations there's value in having some kind of tooling or API that partitions and simulates specific environments for processes running in the same kernel space. For better or for worse, in 2019, Docker is the best solution to this problem.
I think where Docker tends to shine is when you are operating at scale. For larger players using VMs to recreate a computing environment is expensive and eats into your bottom line. In such situations there's value in having some kind of tooling or API that partitions and simulates specific environments for processes running in the same kernel space. For better or for worse, in 2019, Docker is the best solution to this problem.