A technique used for networks which require to be working continually and where the failure of an individual computer or the loss of the data on the computer would adversely affect the network or its applications. There are two main types of replication: data replication where identical data is stored on a number of servers, and computer duplication where a number of computers are executed in parallel, checking each other's results; when one disagrees it is assumed to be malfunctioning and is temporarily retired.