Scholarlink

<p>With an increasing complexity of the submitted tasks by users and the diversity of requirements, many computing platforms have begun to incorporate heterogeneous computing resources to meet user demands, which also brings new challenges to computing platform resource management. Task scheduling, as the core link of resource management, its optimization is particularly crucial. Efficient task scheduling strategies can not only significantly enhance the service experience of users, but also improve economic benefits for cloud service providers. In this paper, the grey wolf optimization (GWO) is improved which names as Improved-GWO; then, a novel hybrid algorithm called HIGWOLS is proposed to address the task scheduling problem for heterogeneous virtual machines. HIGWOLS is formed by combining the Improved-GWO with local search algorithm which improves the convergence speed and solution accuracy of traditional GWO. To validate the proposed hybrid method, a series of experiments are performed using different real world datasets with various sizes. The proposed hybrid method is compared with other swarm intelligent algorithms, such as PGSAO, GA-GWO and whale optimization algorithm (WOA). The proposed hybrid algorithm reduces the makespan by 8.9%, 24.6%, and 17.8% compared to PGSAO, GA-GWO and WOA, while improving resource utilization rate by 10.4%, 22.8%, and 27.5%, and reducing degree of load imbalance by 91.8%, 93.6%, and 90.9%, respectively. The experimental results indicate that the HIGWOLS has significant advantages when using HPC2N and NASA iPSC.</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

A hybrid algorithm for multi-objective task scheduling in heterogeneous cloud computing

  • 期刊论文

  • 发表日期：

    2025年7月11日

查看全文
  • Youli Zhang,
  • Hu Zhang,
  • Changjian Song

摘要

With an increasing complexity of the submitted tasks by users and the diversity of requirements, many computing platforms have begun to incorporate heterogeneous computing resources to meet user demands, which also brings new challenges to computing platform resource management. Task scheduling, as the core link of resource management, its optimization is particularly crucial. Efficient task scheduling strategies can not only significantly enhance the service experience of users, but also improve economic benefits for cloud service providers. In this paper, the grey wolf optimization (GWO) is improved which names as Improved-GWO; then, a novel hybrid algorithm called HIGWOLS is proposed to address the task scheduling problem for heterogeneous virtual machines. HIGWOLS is formed by combining the Improved-GWO with local search algorithm which improves the convergence speed and solution accuracy of traditional GWO. To validate the proposed hybrid method, a series of experiments are performed using different real world datasets with various sizes. The proposed hybrid method is compared with other swarm intelligent algorithms, such as PGSAO, GA-GWO and whale optimization algorithm (WOA). The proposed hybrid algorithm reduces the makespan by 8.9%, 24.6%, and 17.8% compared to PGSAO, GA-GWO and WOA, while improving resource utilization rate by 10.4%, 22.8%, and 27.5%, and reducing degree of load imbalance by 91.8%, 93.6%, and 90.9%, respectively. The experimental results indicate that the HIGWOLS has significant advantages when using HPC2N and NASA iPSC.