In one embodiment, a cluster system comprises a plurality of cluster nodes for executing a plurality of applications, and a management process for controlling the plurality of cluster nodes, wherein the management process is operable to identify a first application that is not meeting a service level objective (SLO), to select a cluster node that is not currently executing the first application, to progressively decrease routing of transactions to the selected cluster node for a second application, and to progressively increase routing of transactions to the selected cluster node for the first application.