Enhancing Energy-Efficiency in Data Centers by Leveraging Computational Steering and Dynamic Task Allocation

Data centers are critical components of the infrastructure of digital services but have a number of problems related to the management of resources, energy consumption, and organizational efficiency. The following paper describes a computational steering approach for resource management and energy consumption in cloud computing data centers. The system combines Round Robin and SJF scheduling with a Random Forest Regressor model for job duration prediction. Performance evaluation demonstrates significant improvements: SJF provided lesser energy utilization where it was brought down to 17.44 kW from 18.83 kW, enhanced CPU utilization from 16.31% to 15.82% and energy consumption variation was lowered to 5.79 from 39.4. Similarly, Round Robin scheduling minimized energy consumption to 17.76 kW, and maximized CPU utilization to 15.64%. Random Forest Regressor model gave an R-squared of 0.98, showing the model's efficiency in estimating job duration. It involves a Streamlit frontend that emulates real-time monitoring and control for scheduling tasks, and shows possible improvements in resource usage and energy consumption. This research work helps to effectively manage data center energy usage and is anchored on United Nation Sustainable Development Goals to provide a practical solution to energy efficient data center management.