Clusters

A cluster is a group of one or two node pairs. Therefore, you can assign up to four SAN Volume Controller nodes to one cluster. All configuration and service is performed at the cluster level. Some service actions can be performed at node level, but all configuration is replicated across all nodes in the cluster. Because configuration is performed at the cluster level, an IP address is assigned to the cluster instead of to each of the nodes.

All your configuration and service actions are performed at the cluster level. Therefore, after configuring your cluster, you can take advantage of the virtualization and the advanced features of the SAN Volume Controller.

Cluster state and the configuration node

The cluster state holds all configuration and internal cluster data for the cluster. This cluster state information is held in nonvolatile memory. If the mainline power fails, the two uninterruptible power supplies maintain the internal power long enough for the cluster state information to be stored on the internal disk drive of each node. The read and write cache information is also held in nonvolatile memory. Similarly, if the power fails to a node, configuration and cache data for that node will be lost and the partner node attempts to flush the cache. The cluster state is still maintained by the other nodes on the cluster.

Figure 16 shows an example cluster containing four nodes. The cluster state shown in the grey box does not actually exist, instead each node holds a copy of the entire cluster state.

The cluster contains a single node that is elected as configuration node. The configuration node can be thought of as the node that controls the updating of cluster state. For example, a user request is made (item 1), that results in a change being made to the configuration. The configuration node controls updates to the cluster (item 2). The configuration node then forwards the change to all nodes (including Node 1), and they all make the state-change at the same point in time (item 3). Using this state-driven model of clustering ensures that all nodes in the cluster know the exact cluster state at any one time.

Figure 16. Cluster, nodes, and cluster state.
Cluster, nodes, and cluster state
Related topics

(C) Copyright International Business Machines Corporation 2003, 2004