MCWDRL: Multi-Cloud Workflow Scheduling Using Deep Reinforcement Learning and Improved Workflow Segmentation for Multi-Cloud Environments

Abstract

The rapid growth of cloud computing has led to complex workflow scheduling challenges in multi-cloud environments, where efficient resource utilization, minimized makespan, and reduced costs are paramount. Traditional scheduling algorithms often struggle to adapt to the dynamic nature of cloud resources and workflows, leading to suboptimal performance. To overcome this issue, the paper proposed a novel Multi-Cloud Workflow Scheduling using Deep Reinforcement Learning and Improved Workflow Segmentation (MCWDRL) approach that addresses these challenges by synergistically combining deep reinforcement learning, workflow segmentation, and scheduling techniques. MCWDRL optimizes workflow execution by partitioning workflows into manageable segments, leveraging deep reinforcement learning to adapt to changing cloud environments, and implementing a scheduling policy that minimizes makespan and resource cost. By integrating these components, MCWDRL offers a robust and adaptive solution for workflow scheduling in multi-cloud environments, outperforming existing algorithms and demonstrating significant improvements in workflow execution efficiency and resource utilization.