Service Level Objective Adaptive Energy Efficiency Management

Cloud applications are transforming to microservice based deployment. Microservice comes with many advantages-scalability, agility, availability – and it suits the complicated cloud applications. For microservice deployment, service quality metrics play important role to ensure reliable user experiences for those applications built upon many small services, and the service level objective (SLO) is adopted to govern resource provision for optimal operation cost. Amid growing awareness on data center carbon emission, more organizations are paying attention to energy-efficient practices in cloud deployment to reduce energy consumption and operational carbon footprint. Modern processors provide fine-grained power scaling capabilities, e.g., Intel Xeon PCPS (PerCore P-States), this paper presented a dynamic energy efficiency management framework to adaptively scale processor core frequency as well as implicated energy consumption in accordance with service quality requirements. In this paper, we discussed hurdles and solutions for applying platform energy intelligence to microservices that are often with intrinsic platform-agnostic. The experiments proved the proposed method can effectively reduce energy consumption with competitive cost benefits without compromising service quality goals.