Consider, for example, a cluster that has a set of disks attached to it. Initially, a small disk array may have been sufficient for the application, however, over time the application requirements may grow and a new, high performance disk array may be required. A new disk drive or logical disk can be exposed to the cluster and a physical disk resource created. This physical disk resource can then be substituted into the application dependency tree using the same procedures described in section Recovering a cluster disk.