Cluster: A network for all occasions

A cluster is a computer network consisting of networked IT machines. Different forms of networking are used depending on the intended use.

A high-availability cluster (HAC) ensures constant availability and a high level of reliability. If a problem occurs in one node of the network, all running services migrate to another node. Both the hardware and software of an HA cluster must be completely free of singular components that could paralyze the system in the event of a failure. A dual design of all components is therefore the minimum rule for HA clusters. The components of such clusters are often geographically separated by several kilometers, so that the system does not fail even in the event of a disaster ('stretched cluster').

Another problem is changing performance requirements. This is where 'load balancing clusters' (LBC) are used, which divide the workload evenly among themselves in the event of increased requirements. Such systems are often installed for cost reasons. Instead of individual expensive high-performance machines, a swarm of networked standard computers divides the changing tasks evenly between them.

High-performance computing clusters (HPC) are used wherever large amounts of data need to be processed. Science is usually the field of application for such a network. Here, a task is divided up among many computers with the help of a decomposition program, each of which takes on subtasks ('jobs'). The speed of the network and the software used are critical factors here. Today, the so-called 'super computers' mostly use an HPC cluster that operates on a Linux basis.