| ABSTRACT |
Deterministic task scheduling in traditional homogeneous cloud data centres is straightforward, but increasing resource heterogeneity has made efficient scheduling more complex. Deterministic methods struggle to meet Quality of Service constraints, as improper VM-to-task mapping can degrade performance, leading to underutilization or overutilization of virtual machines (VMs). To address this challenge, researchers are turning to metaheuristic approaches, such as Marine Predator Algorithm (MPA). This product modifies MPA to use discrete value instead of the original continuous values, and introduces a selective mutation condition to reduce the number of iteration while maintaining a near-optimal result as the original MPA.
|