QWhale and SARSAWhale: Energy-Efficient and Energy-Aware Algorithms for High-Load Cloud Environments

Cloud computing has transformed resource management by providing on-demand access to shared computational resources, yet task scheduling in dynamic environments remains a critical challenge due to fluctuating workloads, varying resource availability, and diverse task priorities. Traditional optimization algorithms often fail to adapt effectively to these dynamic conditions. This paper introduces two novel hybrid frameworks: the Q-Whale Algorithm (QWA) and the SARSA-Whale Algorithm (SWA). Both approaches integrate the Whale Optimization Algorithm (WOA) with reinforcement learning techniques—Q-learning and SARSA, respectively—to address the challenges of real-time task scheduling. QWA combines the global search and exploration capabilities of WOA with the adaptive decision-making of Q-learning, while SWA leverages SARSA’s on-policy learning mechanism for enhanced convergence and decision-making in dynamic settings. Experimental evaluations in simulated cloud environments reveal that both algorithms outperform traditional scheduling methods, with SWA demonstrating marginally better performance in terms of resource utilization, makespan reduction, and adherence to task deadlines. These findings highlight the potential of hybrid intelligent algorithms in advancing the efficiency and reliability of cloud-based task scheduling systems.