Publication

Abstract : The rapid expansion of data centers has led to a substantial increase in energy consumption, prompting a need for more energy-efficient solutions. This paper presents a comprehensive approach to enhancing the energy efficiency of heterogeneous servers by leveraging Amazon Web Services (AWS) and Machine Learning (ML) techniques. Our proposed solution involves developing an ML model that uses a dataset of VM cloud which consists of various parameters like the cloud ID, timestamp, CPU performance, memory usage, network traffic, energy efficiency and so on. The model uses regression algorithms like Linear Regression, Decision Tree Regressor, K-Neighbor Regressor, Polynomial Regression, and AdaBoost Regressor to predict the energy-efficiency of a system based on the relevant features. The model will be evaluated on the basis of Mean Absolute Error, Mean Squared Error, Root Mean Squared Error, R2 Score, and Mean Absolute Percentage Error and Linear Regression model is noted to provide the least error scores. This model helps in reducing server workloads and dynamically manages resource allocation to minimize energy usage while maintaining optimal performance. The developed ML model is then deployed on the AWS EC2 free tier instance by encapsulating the ML models within a Docker container. This study provides a scalable and economical approach to enhance the energy efficiency in the data centers, with a potential impact on the wide range of cloud -based infrastructures.