Cluster Controller

While computer clusters allow for solving difficult computation problems quickly or maintaining high availability, administrators still need some way to control them. In most clustering schemes, one machine is designated as a cluster controller. This machine is responsible for distributing pieces of work to the other machines, handling failover and receiving output from the other machines. The cluster is typically configured in a master-slave arrangement, with the cluster controller serving as the master. The concept, if not the name, is common in clustering systems. In MAAS, it is known as a NodeGroup and in Hadoop it is called a NameNode.