Optimizing workflow scheduling for efficient resource utilization in scalable cloud computing data centers

The growing demand for cloud computing services has led to a rapid expansion of cloud data centers (CDCs), significantly increasing global energy consumption, driven by underutilized physical machines and continuous cooling overhead. To address these challenges, cloud providers are adopting green solutions such as dynamic consolidation, which minimizes the number of active physical machines while maintaining system performance. In this comparative study, we model task arrivals using five probability distributions (Normal, Lévy, Pareto, Chi-square, and Binomial) to explore their impact on the scheduling efficiency of six well-established metaheuristic algorithms; genetic algorithm (GA), ant colony optimization (ACO), cuckoo search (CS), particle swarm optimization (PSO), artificial bee colony (ABC), and simulated annealing (SA). By introducing probabilistic variation in task arrival times, the study examines the sensitivity of these algorithms to dynamic workloads. Using the CloudSim simulator, performance is assessed across small- and large-scale CDC environments based on makespan, energy consumption, and resource utilization. Results reveal that Pareto-distributed task arrivals yield consistently strong performance in small-scale scenarios, while PSO paired with chi-square distributions outperforms others in large-scale settings.