We have done a complete cluster installation. It is not as difficult as it seems. Once we have fixed the variables and decided what hardware requirement is needed, it is just a matter of running a shell script that downloads, installs, and configures Cassandra. Multiple data center setups are equally simple, when you figure out initial tokens. Many variables depend on your particular use case, but if you do not have a particular specification in mind, go with the suggested ones or the default one. It is generally good.
Setting up a cluster is probably a one-time task for an organization. It is likely that your first cluster will be just a couple of nodes. It is equally likely that you will stick with that cluster for first production, or at least till you plan to make the first release with Cassandra in your system. In production, the first couple of things that comes to everyone's mind are whether the software is tuned to perform its best? What happens when things start to break...