Int J Performability Eng ›› 2018, Vol. 14 ›› Issue (9): 1958-1967.doi: 10.23940/ijpe.18.09.p3.19581967

Previous Articles     Next Articles

Autonomic Cloud Resource Allocation Method based on LS-SVM and Virtual Allocation

Chenyang Zhaoa, * and Junling Wangb   

  1. aCollege of Information Science and Engineering, Henan University of Technology, Zhengzhou, 450001, China;
    bCollege of Science, Henan University of Technology, Zhengzhou, 450001, China
  • Revised on ; Accepted on
  • Contact: * E-mail address:

Abstract: Current cloud resource allocation cannot be performed autonomously. When a cloud server overloads, the task queue continues to grow, which leads to delay or failure of task execution. In order to solve this problem, an autonomic cloud resource allocation method is proposed in this paper. For each type of task, Least Squares Support Vector Machine (LS-SVM) is used to predict the number of upcoming tasks in the next period by analyzing a time series of historical task numbers. Meanwhile, the queue lengths of various types of tasks are also periodically monitored during each period. Then, according to the predicted task numbers and the real-time queue lengths, Virtual Allocation (VA) is used to autonomously adjust resource allocation for various types of tasks during the task execution. The experiment shows that LS-SVM prediction is more accurate and VA is more effective, which can improve loads of cloud servers and reduce completion time of tasks.

Key words: cloud computing, resource allocation, LS-SVM, virtual allocation, autonomic