Implementing an intelligent learning-based algorithm for efficient task scheduling in cloud computing environments

Cloud technology provides scaled resources to support real-world applications, which often adopt cloud infrastructure for storage and computing. As cloud platforms enlarge their user numbers, optimizing infrastructure usage and balancing the utility and satisfaction between the service provider and users under the Service Level Agreements (SLAs) becomes crucial. Since the dynamism from millions of user workloads makes task scheduling more challenging, especially in the emergence of energy-saving and real-time Quality of Service (QoS) requirements, traditional heuristic scheduling strategy fails to meet the more accurate and optimized operations. We propose an intelligent learning-based cloud task scheduling (ILbCTS) algorithm that leverages Deep Reinforcement Learning (DRL) technology to address this. The intelligent cloud task scheduling is a dynamic process that continuously adapts the action based on the changing states and the available rewards. The empirical studies with job sets of 1000, 5000 and 10,000 show that the ILbCTS algorithm outperforms the existing task scheduling algorithms, such as PSO, MBO and MOPSO, in terms of execution time, energy conservation and success rate of task scheduling.