A cost-effective strategy for Cloud system maintenance. (February 2017)

• Record Type: Journal Article
• Authors: Li, Xinyi; Qi, Yong; Chen, Pengfei; Fan, Yang
• Is Part Of: Computers & electrical engineering. Volume 58(2017)
• Page Start: 176
• Page End: 189
• Publication Date: 2017-02
• Subjects: Cloud computing -- Availability -- Software maintenance -- Restless multi-armed bandit; Computer engineering -- Periodicals; Electrical engineering -- Periodicals; Electrical engineering -- Data processing -- Periodicals; Ordinateurs -- Conception et construction -- Périodiques; Électrotechnique -- Périodiques; Électrotechnique -- Informatique -- Périodiques; Computer engineering; Electrical engineering; Electrical engineering -- Data processing; Periodicals; Electronic journals; 621.302854
• DOI: 10.1016/j.compeleceng.2016.09.009
• Languages: English
• ISSNs: 0045-7906

Abstract

Highlights: Multi-VM maintenance problem is formulated as a Restless Multi-armed Bandit process. Healthy Indices of service VMs are derived from Lagrangian relaxation of the optimization problem. An index based policy is proposed to schedule multi-VM maintenance in the Cloud. Cost-effective Multi-VM Maintenance Strategy has superiority in reducing maintenance cost and improving system availability given limited backup VMs. Abstract: To guarantee the high availability of modern Cloud-based applications, service backup is commonly used to maintain the failed services. However, an ineffective maintenance strategy based on virtual machine (VM) backup will cause a high budget and low resource utilization. This paper proposes a cost-effective multi-VM maintenance strategy to minimize maintenance cost with limited backup VMs in the Cloud. Different from the single-system maintenance strategies that pursue local optima by considering only one system each time, the proposed strategy considers all the running systems simultaneously from a global perspective. We formalize the multi-system maintenance as a restless multi-armed bandit problem, propose the concept of Healthy Indices of VMs with Lagrangian relaxation and develop an index based policy for maintenance scheduling. Given the limited backup VMs, the experiment results show that the proposed cost-effective multi-VM maintenance strategy significantly outperforms single-system maintenance policies in terms of total maintenance cost and system availability. Graphical abstract: