So we have decided to push the required configuration through Chef into our infrastructure. However, any configuration management system gets true power once it allows us to provision different kinds of machines using a minimal piece of code. Imagine having different types of systems in your infrastructure such as dev machines running over laptops, staging environments that are set up on virtual machines, and production environments running on a beefy hardware. It's quite likely that for one class of application, the configuration for these three different sets of machines might be different. However, in most cases, the amount of difference between these configurations is minimal. For example, let's say you are managing infrastructure, running Hadoop. It's pretty obvious that the amount of data that you'll be working with in a staging environment would be substantially less than in a production environment. This would lead to a difference in the configuration...