The full text of this article
Cloud workflow scheduling algorithm based on reinforcement learning
by Delong Cui; Zhiping Peng; Wende Ke; Xiaoyu Hong; Jinglong Zuo
International Journal of High Performance Computing and Networking (IJHPCN), Vol. 11, No. 3, 2018
Abstract: How to fairly schedule the multiple workflow with multiple priorities submitted at different times has become an increasing concern in workflow management system (WMS). To solve the problem, a novel workflow scheduling algorithm based on reinforcement learning is proposed in this study. In our scheme, we first define some basic concepts of reinforcement learning in cloud computing, such as state space, action space and immediate reward. Then single DAG and multiple DAG cloud workflow scheduling algorithm based on reinforcement learning are designed respectively. Reinforcement learning sets up a policy to maximise the cumulative rewards in the long-term through the repetition of trial-and-error interactions in cloud computing environment. Finally, we analyse algorithm performance by using queuing theory. We use real cloud workflow to test the proposed scheme. Our results, on the one hand, demonstrate the proposed scheme can reasonably schedule multiple DAGs with multiple priorities and improve utilisation rate of resources better and, on the other hand, show optimisation object function achieves fair workflow scheduling in cloud computing environment.
Online publication date: Thu, 10-May-2018
is only available to individual subscribers or to users at subscribing institutions.
Go to Inderscience Online Journals to access the Full Text of this article.
Pay per view:
If you are not a subscriber and you just want to read the full contents of this article, buy online access here.
Complimentary Subscribers, Editors or Members of the Editorial Board of the International Journal of High Performance Computing and Networking (IJHPCN):
Login with your Inderscience username and password:
Want to subscribe?
A subscription gives you complete access to all articles in the current issue, as well as to all articles in the previous three years (where applicable).
See our Orders page to subscribe.
If you still need assistance, please email firstname.lastname@example.org