Reliability-Aware Energy Efficiency in Web Service Provision and Placement

Reliability is a critical concern in the provision and placement of web services. A breakdown of service would seriously reduce customers´ satisfaction, and thus harm the revenue of service providers. To maintain a high reliability, the common approach is deploying multiple service instances across different physical servers. This would inevitably raise another concern of energy consumption. Thus, greening web services also becomes an important issue. In this paper, we study the fundamental tradeoff between reliability and energy consumption, and propose an optimization framework that considers both factors. In specific, we build a continuous-time Markov model to analyze the steady-state reliability and mean time to failure (MTTF) from a service-oriented perspective, and obtain the minimum number of service instances to meet the given reliability requirement. Then, we show that deploying these instances in the server cluster to minimize energy consumption is NP-hard. To this end, we propose a heuristic algorithm to approximate the result. The analytical and experimental results show the effectiveness, and the approximation ratio is less than 1.25 for 90% of the data sets we use.