Fault-Tolerant Scheduling of Scientific Workflow in Cloud Computing

Cloud computing, a pivotal technology offering computational services through diverse deployment models, faces challenges in experimentation and testing due to high infrastructure access costs and limited scalability of testbeds. This research introduces Workflowsim, a simulation tool building on CloudSim, facilitating high-level workflow management. The system in-corporates dynamic clustering, enabling fine-grained task consolidation into coarser clusters. However, existing algorithms, such as MinMin and FCFS, exhibit runtime overhead in layered architectures. The paper proposes fault-tolerant scheduling using Dynamic ReClustering, addressing task and job failures in scientific workflows. Employing a Failure Generator and Failure Monitor, the system simulates failures, adjusting clustering dynamically through Reclustering. The proposed algorithm optimizes task scheduling based on dependencies, significantly reducing waiting times. The experi-mental results, implemented on Workflowsim, demonstrate the effectiveness of the proposed fault-tolerant scheduling algorithm, outperforming existing methods in makespan and mean execution time for scientific workflows.