Appendix C: Reviewing Key Failover Cluster Terms

Applies To: Windows Server 2008

This appendix provides definitions for a set of key failover cluster terms.

For information about creating a design for a failover cluster, see Failover Cluster Design Guide.

Term	Definition
failover cluster	A set of independent computers that work together to increase the availability of services and applications. The clustered servers (called nodes) are connected by physical cables and by software. If one of the nodes fails, another node begins to provide services (a process known as failover).
failover	The process of taking a clustered service or application offline on one node and bringing it back online on another node. When a clustered instance of a service or application goes offline, all the resources that are a part of the clustered instance go offline. The offline and online transitions occur in a predefined order. For example, a dependent resource is taken offline before the resources that it depends on are taken offline. Likewise, a dependent resource is brought online after the resources that it depends on.
failback	The process of returning a clustered service or application to its preferred node after the node has failed and then come back online.
clustered instance	A group of resources in a failover cluster that are handled as a unit of failover. When the resources fail over, they fail over together in a predefined way to another node in the cluster.
Cluster service	The essential software component that controls all aspects of server cluster or failover cluster operation and manages the cluster configuration database. Each node in a failover cluster or server cluster runs one instance of the Cluster service.
quorum	The number of elements that must be online for a given failover cluster to continue running. The elements relevant in this context are nodes or, in some cases, a witness disk or witness file share. Each element included in the quorum, except a witness file share, contains a copy of the cluster configuration. The Cluster service works to keep all copies of the cluster configuration synchronized at all times.
witness disk	A disk in the cluster storage that is designated to hold a copy of the cluster configuration database. A failover cluster has a witness disk only if this is specified as part of the quorum configuration.
witness file share	A file share that is part of a Node and File Share Majority, which is a quorum configuration. The witness file share can act as a tie-breaker, determining whether a given set of nodes can function as a cluster at a given time. A failover cluster has a witness file share only if this is specified as part of the quorum configuration.