Environment data processing for a data centre (1)

• Main Author: Yeo, Haw Lin
• Other Authors: Tan Rui
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2024
• Subjects: Computer and Information Science; Engineering
• Online Access: https://hdl.handle.net/10356/174805
• Institution: Nanyang Technological University

Data centres have aided in the development of this ever-digitalising world due to the increasing need for processing and data storage capacities. The huge increase in the number of data centres worldwide has led to concerns for their impact on the environment due to their significant power consumption. Part of this environmental impact by data centres is contributed by the immense amount of heat energy generated by the data centre servers and cooling systems. Due to these concerns, systems have been developed to capture the heat generated by data centres, and from there, the need for methods to predict the heat generated by the data centres arise. This study uses the Tropical Data Centre (TDC) 1.0 dataset to examine and explore two relationships. The first relationship is that between the power consumed by the data centre racks and the amount of heat they generate. The second relationship is that between the environmental factors and the total heat leaving the data centre racks. This study used various machine learning models to study these relationships, with the first part using Linear Regression, Polynomial Regression, and Neural Networks. The second part used the aforementioned models, along with Classification and Regression Trees (CART), Random Forest, Gradient Boosting, and Support Vector Machines (SVM). The results of the study suggest that there is no clear relationship in the first part of the study that could be explained with the machine learning models used, while the results of the second part of the study provided favourable results that could potentially explain its relationship. This study provides an insight into the possibility of using certain models to accurately predict the total heat leaving the data centre, meaning that environmental variables can possibly be used to enhance the heat capturing systems that data centres use.