A node is a unit in a cluster that has control of its own amount of CPU and memory resources. It also has its own OS. In most implementations of Hadoop, this is a single server; although it is possible to have more than one node on a physical server if each node can control its assigned compute and memory resources (typically through virtualization).