Multi-Dimensional Performance Optimization in Cloud Computing

Abstract

The challenge of multi-dimensional performance optimization has been extensively addressed in the literature based on deterministic parameters. Since resources in Cloud Computing platforms are geographically separated and heterogeneous, it is rather difficult to apply a uniform distribution algorithm for achieving various optimization goals. Based on the analysis of cloud service performance measures, this paper proposes an approach for optimal network resource distribution managed by the multi-agent system (MAS), which is aimed to satisfy both the users’ and the service providers’ requirements. Moreover, a communication algorithm that uses the universal generating function technique is proposed to obtain the service time distribution efficiently.