Notes on the concept of quorum in Windows Server 2008 or Windows Server 2008 R2
Quorum is not a new concept for clustering in Windows Server products, but the implementation, and thus the behavior, was new in Windows Server 2008. The new quorum model can be adapted more to the high availability characteristics requested by the system administrator to support applications, and is less tightly coupled to the way cluster hardware is hooked together. Furthermore, the new quorum model has eliminated the single point of failure that existed in previous clustering releases.
If you choose Node and Disk Majority for the quorum mode, when you select the disk witness, a \Cluster folder is created at the root of the selected disk, and cluster configuration information is stored there. The same information is also stored on each node.
\Cluster folder contains the cluster registry hive
No more checkpoint files or quorum log files
The three main reasons why quorum is important are to ensure consistency, act as a tie-breaker to avoid partitioning, and to ensure cluster responsiveness.
Because the basic idea of a cluster is multiple physical servers acting as a single logical server, a primary requirement for a cluster is that each of the physical servers always has a view of the cluster that is consistent with the other servers. The cluster hive acts as the definitive repository for all configuration information relating to the cluster. In the event that the cluster hive cannot be loaded locally on a node, the Cluster service does not start, because it is not able to guarantee that the physical server meets the requirement of having a view of the cluster that is consistent with the other servers.
A witness resource is used as the tie-breaker to avoid “split” scenarios and to ensure that one, and only one, collection of the members in a distributed system is considered “official.” A split scenario happens when all of the network communication links between two or more cluster nodes fail. In these cases, the cluster may be split into two or more partitions that cannot communicate with each other. Having only one official membership prevents unsynchronized access to data by other partitions (unsynchronized access can cause data corruption). Likewise, having only one official membership prevents clustered services or applications being brought online by two different nodes: only a node in the collection of nodes that has achieved quorum can bring the clustered service or application online.
To ensure responsiveness, the quorum model ensures that whenever the cluster is running, enough members of the distributed system are operational and communicative, and at least one replica of current state can be guaranteed. This means that no additional time is required to bring members into communication or to determine whether a given replica is guaranteed.
