Self-Optimization in Autonomic Systems

Abstract—Autonomic computing is a research area that ex-tends to numerous different fields of science. We describe how autonomic computing can be used to overcome many problems the IT industry is facing today. Autonomic computing systems are, by definition, self-configuring, self-healing, self-optimizing and self-protecting. We show some examples of existing systems that have these self- * capabilities. We look at what utility functions are and how they can be used in self-optimization. We describe a two-level architecture for self-optimization in a data center scenario. The architecture is flexible and suits very different kinds of applications. It supports fair resource allocation in a system where there can be very heterogeneous applications using shared resources. I