Skip to Main Content
Recently a great deal of research has been under-taken in the area of automating the enterprise IT Infrastructure. For enterprises with a large number of computers the IT Infrastructure operation represents a considerable amount of the enterprise budget. Autonomic Computing Systems are systems which were created for minimizing this budget component. They were meant to correct and optimize the IT infrastructure's own self-functioning by executing corrective operations without any need for human interventions. In most cases, where autonomic computing systems have been developed, this was achieved by the addition of external global controllers monitoring the sub-systems of the enterprise IT Infrastructure, determining where changes should be made and applying appropriate commands to implement these changes. Self-Organizing systems on the other hand are systems which reach a global desired state without the use of a central authority which in certain case is the human operator. This paper introduces a general architecture and appropriate algorithms for a self-organizing system which automates a cluster of servers and which maintains an equal desired response time across all the servers. The self-organizing control applies either in the case of homogeneous servers or heterogeneous ones. Furthermore, a simple controller can be built to add or remove servers from the cluster where the controller is itself a peer in the self-organizing system. Simulation data for an autonomic computing system made out of a few cluster servers which are controlled by a self-organizing controller is presented.