Hybrid Meta-heuristic Genetic Algorithm: Differential Evolution Algorithms for Scientific Workflow Scheduling in Heterogeneous Cloud Environment
Hybrid Meta-heuristic Genetic Algorithm: Differential Evolution Algorithms for Scientific Workflow Scheduling in Heterogeneous Cloud Environment
复制标题
混合元启发式遗传算法：异构云环境下科学工作流调度的差分进化算法
DOI：
10.1007/978-3-031-18344-7_2
复制
发表时间：
2022
期刊：
影响因子：
--
通讯作者：
A. Alwan
中科院分区：
文献类型：
--
作者：
Faten A. Saif;R. Latip;M. N. Derahman;A. Alwan
云计算通过互联网提供在线服务的巨大潜力吸引了分布式部门的注意，因为它具有巨大的能力，包括存储，处理，软件，数据库和服务器，这些服务器由地理上分散的远程用户通过互联网同时共享。通过大数据平台生成的大量数据的增加以及通过网络连接的物联网设备的使用利用了云的计算能力。然而，云的高利用率导致特定任务的执行时间更长。本文提出了一种混合的云计算工作流调度策略--差分进化遗传算法（GA-DE）。本研究旨在探讨异构云计算如何影响工作流调度。本研究的目的是减少完工时间和验证元启发式技术是更适合于分布式环境，通过比较它与现有的算法，如HEFT-向下排名，HEFT-向上排名，HEFT-级别排名，和元启发式算法GA。通过与三个科学工作流程（表观基因组学，Cybershake和蒙太奇）进行广泛的实验来验证所提出的算法。通过仿真结果证明了GA-DE算法在最大完工时间方面优于其它同类算法。实验证明，蒙太奇科学工作流更适合于异构云计算环境下的工作流调度。
The gaint cabailities of cloud computing in providing online services via Internet attract the attention of the distributed sector due to its huge abilities that include storage, processing, software, databases, and servers that are shared simultaneously over the Internet by remote users geographically dispersed. Increasing the enormous amount of generating data through big data platforms and the use of IoT devices connected via the network have exploited the computational power of the cloud. However, the high utilization of the cloud leads to a longer execution time for a specific task. This paper proposing the hybrid strategy of scheduling the workflow in cloud computing called Genetic Algorithm with Differential Evolution (GA-DE). This research aims to investigate how heterogeneous cloud computing affects workflow scheduling. This study is aimed at reducing makespan and verifying if the metaheuristic technology is more suitable for the distributed environment by comparing it to existing heuristics, such as HEFT-Downward Rank,HEFT-Upward Rank,HEFT-Level Rank, and meta-heuristic algorithm GA. The proposed algorithm is validated through extensive experiments compared to three scientific workflows (Epigenomics,Cybershake,and Montage). Based on the simulation result GA-DE algorithm proves its superiority against the other comparing algorithms in term of makespan. Furthermore, the conducted experiment proves that montage scientific workflow is more proper for executing workflow scheduling in heterogeneous cloud computing.