首页|Optimized dynamic task scheduling in cloud computing for big data processing
Optimized dynamic task scheduling in cloud computing for big data processing
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NETL
NSTL
Springer Nature
Abstract In the current tech landscape, Big Data processing is necessary due to the heavy reliance on data-driven technologies. Traditional computing struggles with the sheer volume of data, prompting the integration of cloud computing with Big Data to unlock vast possibilities and mitigate data management challenges. Cloud services facilitate resource access, catering to the Big Data processing needs of diverse organizations, from small-scale to large-scale enterprises. In cloud computing, task scheduling plays a pivotal role in resource allocation. Addressing the shortcomings of static task scheduling, this study introduces dynamic task scheduling. This approach leverages a support vector machine (SVM) for VM classifications and an optimized moth flame optimizations (MFO) algorithm for efficient task allocation. The SVM classifies VMs into four categories based on workload status: unstable/high resource utilization, moderately stable/moderate utilization, and two categories of stable VMs with differing resource needs. Subsequently, the MFO algorithm allocates tasks to these VM categories, focusing on enhancing load balancing and system efficiency. Comparison with traditional particle swarm optimizations and min–max algorithms highlights the superiority of the suggested method. It achieves notable improvements: reducing task waiting time (TWT) by 20%, enhancing task finishing time (TFT) by 15%, and boosting resource utilization efficiency by 25%. Consistently outperforming conventional methods across diverse metrics ensures effective task allocation and system optimization for cloud-based Big Data processing. This research introduces an efficient dynamic task scheduling framework, significantly refining resource allocation and system efficiency within cloud environments. The proposed model signifies substantial advancements over traditional algorithms, catering to the demands of modern data-driven technologies.
D. Radhika、M. Duraipandian
展开 >
Vivekanandha College of Engineering for Women, Elayampalayam
Hindusthan Institute of Technology, Malumichampatti
NETL
NSTL
Springer Nature
