A Distributed Autonomic Management Framework for Cloud Computing Orchestration

Abstract

Due to constant workload growth, the infrastructure used to support cloud computing (CC) environments increases in size and complexity. As a consequence of that, human administrators are not able to monitor, analyze, plan and execute actions upon the environment, seeking goals such as the environment optimization and service level agreements fulfillment. This proposal provides an autonomic framework to create virtual machines migrations and heuristics to select hosts to be activated or deactivated when needed. Moreover, the framework proposed in this paper works in a distributed way using multi-agent systems concepts. We provide an architecture to deal with the size, heterogeneity and dynamism of CC environments. Further, our proposal was added to the CloudStack platform as a plug-in for validation and experimentation.