High availability and configuration management, RFC

It kind of looks like the 'guide' skips Chef altogether. Opscode docs make it look pretty easy to scale up, can anyone with a large Chef deployment share what are the pinpoints?