Minimizing Energy Consumption in Data Centers Using Embedded Sensors and Machine Learning

Abstract

   Cloud Data Centers (DCs) consume extensive amounts of energy, making a significant contribution to environmental concerns. Moreover, with the emergence of 5G and future B5G networks, which are increasingly inclined towards software orientation and reliant on cloud computing, there is an urgent requirement for optimizing the energy consumption of DCs. Therefore, we address this issue by proposing an energy-aware Virtual Machine (VM) placement solution for energy minimization.

In the first part of this study, we propose a highly accurate model for predicting the dynamic power consumption of cloud computing devices. Our proposal takes advantage of the various sensors which are now embedded in physical machines, or more generally in cloud server machines, as well as Performance Monitoring Counters (PMCs) to implement a highly accurate Machine Learning (ML) power prediction model. The core part of this study then integrates the novel feature space of real-time sensors’ measurements and the predictive power model to propose a scalable placement algorithm, enabling proactive and

energy-aware Virtual Machine placements. In addition, it utilizes a new set of temperature-related features that enables proactive hotspot avoidance.

Our ML predictive models, as well as our proposed placement algorithm, were extensively evaluated on a cluster of real physical machines and demonstrated a significantly higher performance as compared to the implemented reference models and algorithms, reducing energy consumption by up to 7%, CPU temperature by 2%, and overloading by 28%.