Deadline‐Constrained Tasks Allocation Through Resubmission and Preemptive Migration in Heterogeneous Cloud Environments

With the increase in resource heterogeneity and dynamic workloads, cloud environments have become more complex, making it challenging to efficiently manage diverse resources. Limited availability and improper allocation often result in task failures, particularly when systems lack the capacity to meet strict deadlines. To address this challenge, effective task allocation algorithms with integrated fault tolerance are essential. This paper presents DTRM that is, Deadline Constrained Task Allocation using Resubmission and preemptive Migration, a novel task allocation algorithm for heterogeneous cloud environments, designed specifically for deadline‐constrained independent tasks. Unlike existing approaches, DTRM introduces a hybrid fault‐tolerant strategy that combines reactive resubmission and proactive preemptive migration, thereby minimizing task failures under deadline constraints. The algorithm operates in three phases: (i) classification of tasks and VMs according to task length and computation power, (ii) prioritization of tasks based on execution time and deadlines, and (iii) allocation to the most suitable VM category. By aligning task requirements with VM capabilities, DTRM achieves better utilization and reduces task rejection. Simulation results on varying task sizes and VM counts demonstrate that DTRM consistently outperforms baseline algorithms such as FCFS, SJF, Priority‐based scheduling, DyMaxMin, RADL, and FTTA. Specifically, DTRM significantly improves makespan reduction, task acceptance, speedup, and efficiency, validating its potential for real‐world deployment in large‐scale cloud environments where meeting deadlines is critical.