当前位置: X-MOL 学术Cluster Comput. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
An efficient load balancing technique for task scheduling in heterogeneous cloud environment
Cluster Computing ( IF 3.6 ) Pub Date : 2021-06-18 , DOI: 10.1007/s10586-021-03334-z
Hadeer Mahmoud , Mostafa Thabet , Mohamed H. Khafagy , Fatma A. Omara

Cloud computing will continue to play a critical role due to its ability to deliver various types of on-demand information technology (IT) services according to customer needs. Unfortunately, cloud computing has faced many challenges. The task scheduling problem is considered one of the main challenges because the user has to pay for a resource. Therefore, a good mapping of users’ tasks on the available resources is required to reduce the total execution time of the users’ tasks (i.e., reduce makespan), execution cost, enhance load balancing, increase resource utilization, etc. Many heuristic algorithms have been existed to solve task scheduling problems. Although, the Heterogeneous Earliest Finish Time (HEFT) heuristic algorithm is considered one of the best task scheduling algorithms in a heterogeneous environment, it does not concern load balancing. In this paper, a modification has been introduced to the HEFT algorithm to satisfy load balancing. The modified algorithm is called Load Balancing HEFT (LB-HEFT). A comparative study has been done among the proposed LB-HEFT, the Enhancement Heterogeneous Earliest Finish Time (E-HEFT), and HEFT algorithms to evaluate the performance of the proposed LB-HEFT algorithm. According to the comparative results, it is found that the proposed LB-HEFT algorithm outperforms the existing E-HEFT and HEFT algorithms by improving load balancing by 43.49% and 72.59%, respectively in average, enhancing resource utilization by 2.28% and 5.61%, respectively in average, and reducing makespan by 7.55%, and 3.75% respectively in average.



中文翻译:

一种异构云环境下任务调度的高效负载均衡技术

云计算将继续发挥关键作用,因为它能够根据客户需求提供各种类型的按需信息技术 (IT) 服务。不幸的是,云计算面临许多挑战。任务调度问题被认为是主要挑战之一,因为用户必须为资源付费。因此,需要用户任务在可用资源上的良好映射,以减少用户任务的总执行时间(即减少makespan)、执行成本、增强负载平衡、提高资源利用率等。许多启发式算法具有为解决任务调度问题而存在。虽然,异构最早完成时间(HEFT)启发式算法被认为是异构环境中最好的任务调度算法之一,它不涉及负载平衡。在本文中,对 HEFT 算法进行了修改以满足负载平衡。修改后的算法称为负载平衡 HEFT (LB-HEFT)。对所提出的 LB-HEFT、增强异构最早完成时间 (E-HEFT) 和 HEFT 算法进行了比较研究,以评估所提出的 LB-HEFT 算法的性能。根据对比结果发现,所提出的LB-HEFT算法优于现有的E-HEFT和HEFT算法,负载均衡平均分别提高43.49%和72.59%,资源利用率提高2.28%和5.61%,分别平均,平均分别减少7.55%和3.75%的完工时间。对 HEFT 算法进行了修改以满足负载平衡。修改后的算法称为负载平衡 HEFT (LB-HEFT)。对所提出的 LB-HEFT、增强异构最早完成时间 (E-HEFT) 和 HEFT 算法进行了比较研究,以评估所提出的 LB-HEFT 算法的性能。根据对比结果发现,所提出的LB-HEFT算法优于现有的E-HEFT和HEFT算法,负载均衡平均分别提高43.49%和72.59%,资源利用率提高2.28%和5.61%,分别平均,平均分别减少7.55%和3.75%的完工时间。对 HEFT 算法进行了修改以满足负载平衡。修改后的算法称为负载平衡 HEFT (LB-HEFT)。对所提出的 LB-HEFT、增强异构最早完成时间 (E-HEFT) 和 HEFT 算法进行了比较研究,以评估所提出的 LB-HEFT 算法的性能。根据对比结果发现,所提出的LB-HEFT算法优于现有的E-HEFT和HEFT算法,负载均衡平均分别提高43.49%和72.59%,资源利用率提高2.28%和5.61%,分别平均,平均分别减少7.55%和3.75%的完工时间。增强异构最早完成时间 (E-HEFT) 和 HEFT 算法来评估所提出的 LB-HEFT 算法的性能。根据对比结果发现,所提出的LB-HEFT算法优于现有的E-HEFT和HEFT算法,负载均衡平均分别提高43.49%和72.59%,资源利用率提高2.28%和5.61%,分别平均,平均分别减少7.55%和3.75%的完工时间。增强异构最早完成时间 (E-HEFT) 和 HEFT 算法来评估所提出的 LB-HEFT 算法的性能。根据对比结果发现,所提出的LB-HEFT算法优于现有的E-HEFT和HEFT算法,负载均衡平均分别提高43.49%和72.59%,资源利用率提高2.28%和5.61%,分别平均,平均分别减少7.55%和3.75%的完工时间。

更新日期:2021-06-18
down
wechat
bug