random yet consistent datasets

I have limited access to real data, and I need some data for quick prototyping, one way to get it is to randomly generate it. And it works well for simple datasets. But once a dataset has internal inter-dependencies that need to check out and stay consistent it all turns south. I am sure I am not the first one who raised this question. So what is my best course of action (algorithms, frameworks, methodologies) to generating consistent yet random datasets?