Task Scheduling Algorithm Based on Virtual Machine Availability Awareness in Cloud Platform

Zhixin Li, Lei Liu, Zeyu Tong · Journal of Engineering Science and Technology Review · 2018

Virtual machine (VM) availability awareness is a key task scheduling technology in a cloud platform.However, the dynamic change and uncertainty of VM availability constitute difficulties for task scheduling, and quality requirements of task services cannot be satisfied and thus seriously affect the task scheduling capacity of the cloud platform.A task scheduling algorithm based on VM availability awareness was proposed to solve the unmatching problem between VM availability and quality of service (QoS) to improve task scheduling capacity of VMs.This algorithm combined available task processing capacities of VMs and task requirement features to establish a differential entropy model between VM availability and task availability requirement.Task availability matching and scheduling was realized through the principle of maximum entropy, and task scheduling was optimized from the aspect of balance of server workloads.Finally, a comparative verification between the task scheduling algorithm and Random and Minmin algorithms was implemented.Results demonstrate that through the task scheduling algorithm based on VM availability awareness, the task execution speed of VMs is higher than those of Random and Minmin algorithms by 28% and 6%, respectively.Therefore, task execution speed of the cloud platform is significantly elevated, and server workloads are more balanced than those in Random and Minmin algorithms.Within the same time, QoS satisfaction rate is higher than those of Random and Minmin algorithms by 10% and 2%, respectively.QoS task requirement is satisfied while task completion time is reduced.This study concludes that the VM availability awareness method satisfies the task requirements and improves task processing performance of the VM in the cloud platform.Relevant conclusions can provide technical support for task scheduling of the VM in the cloud platform.

Read the paper · More papers on PaperTik