An Intelligent VM Allocation Framework for Handling High-Priority Jobs in Cloud Computing: A Dynamic Load Balancing Approach

Abstract

Efficient resource allocation in cloud computing is essential for managing workloads with varying priorities and deadlines. This research proposes an intelligent Virtual Machine (VM) allocation mechanism that dynamically prioritizes tasks based on job deadlines and characteristics. The framework ensures that high-priority jobs (with tight deadlines) are allocated resources immediately, even if it requires preempting low-priority jobs (with relaxed deadlines). The preemption process evaluates the lease type of running low-priority jobs, prioritizing cancellable jobs for termination and suspendable jobs for pausing, while non-preemptable jobs remain unaffected. Suspendable jobs with the least progress are prioritized for preemption to minimize resource wastage. Once a high-priority job completes, paused suspendable jobs are resumed based on resource availability and deadlines, while cancellable jobs are terminated permanently. This approach optimizes resource utilization, reduces response times for high-priority tasks, and ensures efficient load balancing in dynamic cloud environments. The proposed mechanism enhances the performance and reliability of cloud systems in handling complex workloads with diverse priorities and deadlines.