It does all of that. The structure of the system is such that you can call any function on any node in the cluster. In order to do it well, you need to unders

It does all of that. The structure of the system is such that you can call any function on any node in the cluster. In order to do it well, you need to understand your own abstractions. If we have a cluster are we better off passing every function around the cluster or running some locally? Is the network overhead worth it in every case or only in some? It's up to your code to make those determinations about your application, but everything you describe is fairly simple to implement on the BEAM.