Computercluster

A computer cluster is a set of interconnected computers that work together to perform tasks as a single system. Each node contributes processing power, memory, and storage, while cluster software coordinates execution, reliability, and resource sharing.

Nodes are linked by high-speed networks such as Ethernet or InfiniBand. A typical cluster includes compute nodes, a master or head node that schedules jobs, and shared storage. Interconnect topology, like fat-tree or mesh, affects performance and resilience.

Common types include compute (HPC) clusters for scientific simulations using MPI, high-availability clusters that provide failover, load-balancing clusters for web services, and storage clusters for scalable data access.

Software and management involve job schedulers (Slurm, PBS, Grid Engine), parallel programming interfaces (MPI, OpenMP), and often virtualization or containers. Monitoring and configuration tools help operators track performance and maintain reliability.

Advantages are increased performance, scalability, and fault tolerance; disadvantages include cost, complexity, and maintenance effort. Clusters are used in research, data analytics, rendering, and enterprise web services, and may operate on-premises or in hybrid environments with cloud resources.

Origin traces to Beowulf clusters in the 1990s, using commodity hardware for parallel processing. Since then, advances in interconnects, software, and accelerators have made clusters central to modern high-performance computing and data centers.