Clustering Fundamentals

Suppose you have configured a two-node cluster, with a web server installed on each of the nodes. Each of the servers in the cluster hosts two websites. All the content for all four websites is stored on a shared disk subsystem connected to each of the servers in the cluster. Figure 9.1 shows how such an environment might look.

Figure 9.1. The basic cluster architecture.

During normal operation, each clustered node is in constant communication with the other nodes in the cluster through periodic polling. In this way, a node can quickly detect whether something happens to another node in the cluster.

If node 2 fails due to some hardware or software problems, users currently attached to the web server will lose their connections. The IP address associated with node 2 and all its services are migrated to node 1. Users would likely have to reload their web pages, which would be available from the new node within a few seconds.

After the problem in node 2 is located and repaired, it is restarted and automatically re-inserts itself into the cluster. Node 1 detects the return of node 2 and seamlessly passes back all the addresses and services originally assigned to node 2. The cluster returns to its normal configuration without any administrator intervention.