Scholarlink

<p>In cloud computing, effective task scheduling is crucial for improving resource use, reducing execution time, and ensuring a balanced workload across virtual systems. However, the dynamic and varied nature of cloud environments makes traditional scheduling methods inadequate for meeting multiple goals. To tackle these challenges, this paper presents CloudSched-GA, an adaptive genetic algorithm (GA)-based scheduling framework. The algorithm adjusts its crossover and mutation strategies to avoid early convergence and maintain diversity in solutions. A multi-objective fitness function is used to optimize makespan, load balance, and execution cost at the same time. Experimental results on a synthetic dataset of 200 tasks and 5 different virtual machines indicate that CloudSched-GA reduces makespan by 40%, improves load balancing by 75%, and achieves a score up to 40% lower in combined fitness compared to traditional methods like Shortest Job First (SJF) and First-Come-First-Serve (FCFS). These results show that the proposed algorithm is scalable, efficient, and robust, making it suitable for real-world cloud scheduling applications.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

CloudSched-GA: an adaptive genetic optimizer for efficient and balanced task scheduling in cloud ecosystems

  • 期刊论文

  • 发表日期：

    2025年10月15日

查看全文
  • Fatma M. Talaat,
  • Alyaa A. Hamza

摘要

In cloud computing, effective task scheduling is crucial for improving resource use, reducing execution time, and ensuring a balanced workload across virtual systems. However, the dynamic and varied nature of cloud environments makes traditional scheduling methods inadequate for meeting multiple goals. To tackle these challenges, this paper presents CloudSched-GA, an adaptive genetic algorithm (GA)-based scheduling framework. The algorithm adjusts its crossover and mutation strategies to avoid early convergence and maintain diversity in solutions. A multi-objective fitness function is used to optimize makespan, load balance, and execution cost at the same time. Experimental results on a synthetic dataset of 200 tasks and 5 different virtual machines indicate that CloudSched-GA reduces makespan by 40%, improves load balancing by 75%, and achieves a score up to 40% lower in combined fitness compared to traditional methods like Shortest Job First (SJF) and First-Come-First-Serve (FCFS). These results show that the proposed algorithm is scalable, efficient, and robust, making it suitable for real-world cloud scheduling applications.