A computer cluster is a collective of individual computers yoked together so they function like a single entity for a variety of tasks. Using clusters can offer a number of advantages for certain kinds of activities. They are often harnessed for scientific research, complex calculations in numerous fields, and cost-effective supercomputing in environments like finance. Computer cluster design requires input from skilled computer scientists who can also maintain the cluster to keep it at peak performance. The types of computers used can determine how robust the computer cluster is in operation, and technicians can also adjust operating systems, applications, and other specifications to meet specific needs.
Often the computers are identical or close to it, and they are very closely networked with their own dedicated connection. The security for a computer cluster requires a high degree of trust between units to allow them to communicate rapidly, but the cluster can be protected by security measures to prevent intrusion from the outside. For convenience, they are typically physically located together. Technicians can optimize them to perform particular kinds of tasks.
Groups of cluster computers can be cheaper to install and maintain than a supercomputer with similar performance characteristics. For this reason, facilities that need supercomputing capacity may consider a computer cluster. The computers can perform complex calculations including those used in scientific modeling and processing of detailed and challenging equations. They can also balance loads to avoid slowdowns and shutdowns.
Another advantage can be increased stability. If one computer fails, the rest of the cluster can keep operating. This can be important for data integrity or other operations where using a single computer could expose users to the risk of data loss or other issues. Cluster computing can offer robust options for the management of servers, networks, and other complex systems that cannot go down without incurring great expense and potential nuisances like breakdowns in traffic light systems or air traffic control.
Time on a computer cluster may need to be booked by personnel who want access to the system. This allows technicians to schedule usage as well as maintenance and other activities to keep the computers running smoothly. In an emergency, it may be possible to rearrange the schedule to meet the need of someone with an urgent computing need, like rapid modeling to track an evolving storm system. Individual facilities with critical needs may develop their own computer cluster systems to avoid reliance on a system that they do not control.