The cluster consists of one or more computers/machines working together to provide high availability, reliability, and scalability towards service being provided to clients. If one server/machine fails then work/resources get distributed among other machines in the same cluster.
Single node cluster or pseudo-distributed cluster is the one in which all daemon like NameNode, data node, Jobtracker, and tasktracker runs on the single machine. Default replication factor is 1. Multinode cluster is a cluster which is basically used in master-slave fashion where master and slaves runs on different machines and master node/machine runs Namenode and TaskTracker daemons and slave machine runs Datanode and JobTracker daemons.