How do you write a testing framework/system for evaluating distributed algorithms like this? Doing that in a way that's easy to automate, set up, evaluate, and

How do you write a testing framework/system for evaluating distributed algorithms like this? Doing that in a way that's easy to automate, set up, evaluate, and tear down various scenarios seems like the hardest part to me.