Example, Failover Cluster with Multiple Services and Applications

Applies To: Windows Server 2008

In this example, the fictitious company A. Datum needs to run two client-server applications to support company operations. One is a database application that can be partitioned across several servers that coordinate their activity, with one server covering part of the alphabet (records for A-L), and another covering the rest of the alphabet (records for M-Z). The other application is standalone. A. Datum Corporation wants to provide 99.99% availability (down for less than 1 hour per year) for all these applications. These applications are significantly more important to A. Datum than other client-server applications used at the company.

For examples that illustrate other designs, see Evaluating Failover Cluster Design Examples.

Multiple clustered applications before failover

When the failover cluster begins providing service, each node is supporting either an application partition or the standalone application, as shown in the diagram. As indicated by the boxes behind each server, the servers are running at approximately half their working capacity. That is, each server could double its current load and still function adequately.

Multiple clustered applications after failover

The following diagram shows the same design when the application partition on Node 1 has failed over to Node 2.

The two application partitions are still running as separate entities, so that either partition can easily fail back to Node 1 when that node is running again. Each node can adequately handle the application partitions or application running on that node. However, it can be seen that if another node failed, the node that remained would be overloaded. For most three-node clusters, this is not relevant, though, because two nodes must be running for the cluster to function. This is because the cluster is using the Node Majority quorum option, the quorum option that works best for an odd number of nodes. This quorum option simply means that a majority of the nodes must be available and in communication for the cluster to function. The following diagram illustrates this.

In other words, the minimum number of nodes required to keep the cluster functioning (two) is enough to handle the total load of the applications on the cluster. For more information and diagrams illustrating quorum options, see Appendix F: Reviewing Quorum Configuration Options for a Failover Cluster.

Additional references