Modeling, Analysis and Optimization of Dependability-Aware Energy Efficiency in Services Computing Systems

Besides performance, dependability and energy efficiency are two critical concerns during the design, development and management of large-scale services computing systems. In this paper, we jointly consider the performance, dependability and energy efficiency, and optimize the dependability-aware energy efficiency of services computing systems by maximizing the quality of service and dependability revenue and minimizing energy costs. Markov reward models are put forward, and quantitative analysis of them is carried out. In addition, the methodologies for hierarchical model composition and state aggregation are proposed. Furthermore, the optimization problem is formulated as an average reward criterion Markov decision problem, and the algorithm to solve it is introduced. Finally, the LANL service systems are analyzed and optimized as a case study to illuminate how this approach can apply to large-scale systems in reality.