Hardware clustering (sometimes called operating system clustering) is a hardware-based method of turning multiple servers into a cluster (a group of servers that acts like a single system). As a rule, a hardware cluster is created by installing a number of blade servers on the machine that will control the cluster. Each of the blade servers functions independently of the others, although they all respond to the same requests. The operating system of the controlling server is responsible for monitoring the cluster and performing administrative tasks, such as deciding when failover is necessary and assigning the load of a failed node to a functioning server.
A hardware cluster may be active-passive, in which case some redundant servers are reserved for failover duty and do not run any applications of their own. It can also be active-active, in which case all servers in the cluster run their own applications but also reserve resources to allow them to perform failover duty for each other.
Hardware clustering typically requires some software application changes. The alternative to hardware clustering, application clustering, is a fully software-based method of clustering servers. Because application clustering doesn't require specialized hardware, it is usually less expensive to implement than hardware clustering.