Efficient VM migrations using forecasting techniques in cloud computing: a comprehensive review
- PDF / 3,167,261 Bytes
- 30 Pages / 595.276 x 790.866 pts Page_size
- 50 Downloads / 245 Views
(0123456789().,-volV)(0123456789().,-volV)
Efficient VM migrations using forecasting techniques in cloud computing: a comprehensive review Mohammad Masdari1 • Hemn Khezri2 Received: 17 July 2019 / Revised: 13 November 2019 / Accepted: 19 December 2019 Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract High cost of data centers’ energy consumption and its environmental effects such as CO2 emissions have inspired numerous researches to provide more efficient VM management approaches. VM migration is one of the critical VM management activities whose performance has a direct effect on the energy efficiency of cloud data centers (DCs). To conduct a more effective migration process and reduce the number of VM migrations, some of the VM management frameworks apply prediction algorithms to forecast various migration and VM-related factors. This paper presents an extensive survey and taxonomy of the predictive VM migration approaches adapted for the cloud DCs. For this purpose, it first provides the key issues regarding the VM migration and then classifies them based on their applied prediction algorithm. It illustrates the main contributions of each scheme and describes how prediction methods are integrated into the VM migration process, to make it more effective. Moreover, a comparison of the predictive migration schemes is provided. Finally, the concluding remarks and future research areas are specified. Keywords Virtual machine Migration Regression Markov model ANN ARIMA
1 Introduction Cloud computing is an interesting technology aimed to present unlimited virtual resources to remote customers, based on the pay for use model. Various types of cloud computing such as mobile clouds, hybrid clouds, and cloud federations are designed and provided to deal with users’ various functional and non-functional requirements [1-4]. Effective resource management in the cloud data centers (DCs) has a direct effect on their energy efficiency, scalability, performance, and reliability. Moreover, it can reduce the costs incurred to the cloud users and increase the cloud service providers’ (CSPs) profit. Figure 1 indicates the energy-consuming units in the cloud DCs [5]. As shown in this figure, servers or PMs consume much of the & Hemn Khezri [email protected] Mohammad Masdari [email protected] 1
Computer Engineering Department, Urmia Branch, Islamic Azad University, Urmia, Iran
2
Afagh Higher Education Institute, Urmia, Iran
energy in a DC. Reducing power consumption by optimal resource management is very challenging in green cloud DCs. Consequently, a lot of attention has been paid for dealing with the energy efficiency of PMs. Virtualization is one of the baseline techniques used to improve resource management in the cloud DCs. It applies a software layer denoted as the hypervisor, VMM or virtual machine monitor to manage several virtual machines (VMs) on each server or physical machine (PM) [6]. In the virtualized cloud DCs, two common problems affecting the power consumption ar
Data Loading...
