Robustness of Workload Forecasting Models in Cloud Data Centers

This research investigates the performance and robustness of various workload forecasting models within cloud data centers. Workload prediction plays a critical role in resource management, capacity planning, and cost optimization. Cloud service providers rely on accurate workload forecasting models to efficiently allocate resources and minimize costs while ensuring service availability. This study uses a dataset comprising key metrics such as CPU usage, memory usage, disk throughput, and network throughput in cloud environments. The primary objective is to predict CPU usage percentage using machine learning techniques. Four models are employed for prediction: Random Forest Regressor, XGBoost Regressor, AdaBoost Regressor, and LightGBM Regressor. The models are evaluated based on performance metrics like accuracy, mean squared error (MSE), and R-squared (R²). The research aims to identify the most effective algorithm for workload forecasting in cloud data centers and contribute to improving resource management strategies. Through the use of these models, the study demonstrates their strengths and weaknesses in handling complex data patterns. The project provides a foundation for future advancements in workload forecasting models for cloud computing systems.

Keywords: Cloud Data Centers, Workload Forecasting, Resource Management, CPU Usage Prediction, Machine Learning, Random Forest, XGBoost, AdaBoost, LightGBM, Capacity Planning.