A Robust Autonomic Computing Architecture for Server Virtualization

The increasing heterogeneity and complexity of the Information Technology infrastructure encompassing hardware and software applications, services and networks, led to complex, unmanageable and insecure systems. Correspondingly, new approaches in regards to the IT infrastructure automation and operation have been investigated. Autonomic computing has been born and successfully launched as a research area in response to the "must have" solutions for the problems above. It approaches subjects such as self-configuration, self- protection, self-healing, and self-optimization. Its ambitious goals in making a computer environment behaving in a self controlled and intelligently guided manner, lead to various approaches such as intelligent systems, control, learning machines and others. In this paper, a special adaptive technique will be used for the implementation of self-provisioning and self- optimization techniques for autonomic computing with applications to server virtualization. The autonomic computing processes are considered as nonlinear uncertain systems with constant linearly parameterized uncertainty. The approach is motivated by the fact that a complex computer system behaves in time in a nonlinear fashion, its parameters having a stochastic variation, and components of the autonomic computing environment might fail from time to time. The architecture of an adaptive robust autonomic computing environment and its implementation specific components is described. The application domain of the autonomic computing technology has been chosen to be a virtual server environment. An application of this autonomic computing solution to server virtualization is presented at the end of this paper.