A server cluster is a group of redundantly configured servers that work together to provide highly available access for clients to important applications, services, and data while reducing unscheduled outages. The applications, services, and data are configured as cluster resources that can be failed over or cluster migrated between servers in the cluster. For example, when a failure occurs on one node of the cluster, the clustering software gracefully relocates its resources and current sessions to another server in the cluster. Clients connect to the cluster instead of an individual server, so users are not aware of which server is actively providing the service or data. In most cases, users are able to continue their sessions without interruption.
Each server in the cluster runs the same operating system and applications that are needed to provide the application, service, or data resources to clients. Shared devices are connected to and mounted on only one server at a time. Clustering software monitors the health of each of the member servers by listening for its heartbeat, a simple message that lets the others know it is alive.
The cluster’s virtual server provides a single point for accessing, configuring, and managing the cluster servers and resources. The virtual identity is bound to the cluster’s master node and remains with the master node regardless of which member server acts the master node. The master server also keeps information about each of the member servers and the resources they are running. If the master server fails, the control duties are passed to another server in the cluster.