A computer cluster is one where a number of computers are linked together so as to function almost like a single unit. Now, I say almost like a single unit because, they are separate entities and function separately but when accessed from outside seems like a single entity. These individual computers are connected together with a hi-speed LAN (Local Area Network). This task achieved with the help of clustering software which is used to configure these individual systems so that they appear as a single entity for the outer world.

When a set of server machines are connected together in a similar fashion as mentioned above, it is known as a Server Cluster. This technology has redefined the way servers were configured. The central idea behind this whole server clustering was that one single cluster of machines is better than multiple independent machines. The capacity of the server cluster is far more powerful than any of the servers considered individually. Also, this will reduce th cost of hosting if you are hosting multiple servers. If you are planning to host 10 individual servers, hosting a cluster of 6-7 machines might provide the same performance with saving the extra hardware cost. There is the software cost for this clustering, but the efficiency justifies it.

A server cluster will function just like an individual server. When there is a request from one of the client machines, that request is assigned to a machine in the cluster which will be free next. This operation of assigning the activities and managing the cluster is done by a software program which is written on communication libraries. There are different kinds of server clustering like High availability cluster, load balancing cluster, failover cluster among others.

A High Availability cluster is one where there are back up nodes in the cluster which will improve the services when there is an overload on other nodes. A load balancing cluster is one where the extra load in evenly split and assigned to the backup nodes. A failover cluster in one where when a node crashes, all the resources and its work load is assigned to remaining nodes. The last category is the most frequently used server cluster.

While each server in the cluster has its own functions, they often communicate with other nodes to process requests. This necessarily establishes some amount of dependency on each node and the degree of dependency is decided based on how much each node is related to the other one. If all the nodes are identical, then the advantage is that if one server fails, rest of them will take up the load and the services you host are still available. If they are not completely identical, then only partial services of that server will be available.

Having talked about server clustering, the service providers for this are mostly just Linux and Microsoft. Each of them specializes in certain kind of clustering. So, choose the right one based on our requirement and host you services.