Abstract

High-Performance Computing (HPC) environments increasingly rely on virtualization technologies to improve resource utilization, workload isolation, and system flexibility. Virtual Machine (VM) scheduling plays a critical role in determining the performance efficiency of HPC systems, as it governs the allocation of computational resources such as CPU cores, memory, and I/O bandwidth among competing workloads. Inefficient scheduling can lead to resource contention, increased execution time, and underutilization of expensive HPC infrastructure. This paper presents an analytical modeling approach for evaluating and comparing VM scheduling algorithms in HPC environments. The proposed analytical models capture key system parameters, including workload characteristics, resource availability, virtualization overhead, and scheduling policies. By formulating mathematical representations of scheduling behavior, the models enable performance prediction and scalability analysis without the need for extensive simulation or empirical experimentation. The study examines common VM scheduling strategies such as time-sharing, space-sharing, and priority-based scheduling, and evaluates their impact on performance metrics such as turnaround time, throughput, and resource utilization. The analytical framework provides insights into the trade-offs between fairness and performance, and supports informed decision-making in scheduler design and configuration. The results demonstrate that analytical modeling is an effective and scalable tool for understanding VM scheduling dynamics in modern HPC environments