期刊
ARABIAN JOURNAL FOR SCIENCE AND ENGINEERING
卷 47, 期 2, 页码 1821-1830出版社
SPRINGER HEIDELBERG
DOI: 10.1007/s13369-021-06076-7
关键词
Cloud computing; Task scheduling; Cat swarm optimization; Total power cost at datacenters; Energy consumption; PSO-Particle Swarm Optimization; CSO-Cat Swarm Optimization; CS-Cuckoo Search
Efficient task scheduling in cloud computing is crucial in minimizing completion time and maximizing resource utilization. This paper introduces Cat Swarm Optimization algorithm for task scheduling, showing significant improvements in completion time, energy consumption, and total power cost over existing algorithms when applied to HPC2N and NASA workloads.
Efficient Scheduling of tasks is essential in cloud computing to provision the virtual resources to the tasks, effectively by minimizing makespan and maximizing resource utilization in cloud computing. Existing scheduling algorithms talks about makespan and resource utilization, but very few authors addressed the issues named as migration time, energy consumption, total power cost in datacenters. These three mentioned metrics are essential in the view of cloud provider, as by minimizing migration time, energy consumption and total power cost in datacenters cloud provider will be directly benefited. This paper introduces task scheduling by using Cat Swarm Optimization algorithm, which addresses the parameters makespan, migration time, Energy Consumption and Total Power Cost at Datacenters. Scheduling of tasks were done by calculating priorities of tasks at task level, and calculating priorities of VMs at VM level to schedule appropriate mapping of tasks onto VMs. It is implemented by using cloudsim simulator and input to the algorithm is generated randomly from the cloudsim for total power cost, we have used HPC2N and NASA workloads, which are parallel workload archives, which were given as an input to the algorithm. Proposed algorithm is compared against existing algorithms like PSO and CS. From the simulation results, we have observed that we got a significant improvement in different parameters when we used HPC2N and NASA workloads. Makespan is improved by 16%, 10%, 27%, 20% by using HPC2N and NASA workload over PSO and CS algorithms, respectively. Energy Consumption is minimized by 22%, 12%, 31%, 21% by using HPC2N and NASA workload over PSO and CS algorithms, respectively. Total Power cost is minimized by 31% and 39% over PSO and CS algorithms, respectively. Migration time is minimized by 34%, 29%, 20%, 14% by using HPC2N and NASA workloads over PSO and CS algorithms, respectively.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据