As seen in Build in building a Virtualized virtualized 5-Node Hadoop node HDP 2.0 Clustercluster (all within a mac) it is (relatively) easy to build a full-featured multi-node Hadoop cluster using virtualization technologies such as VirtualBox. Obviously, I choose to install Hortonworks Data Platform (HDP) when I'm doing such an activity and I also leverage Ambari. With my setup all my my nodes have access to the internet which lets each connect to the Hortonworks Public Repo when it needs it, but with multiple machines all pulling down the same binaries (all through the same physical pipe – often a wifi signal on my Mac) that process is often pretty slow.
...