|
Title:
|
STABILITY-BASED LIVE MIGRATION USING DYNAMIC MARGIN |
|
Author(s):
|
Ryu Sakamoto, Ryuichi Takahashi, Yoshiaki Fukazawa |
|
ISBN:
|
978-989-8533-84-5 |
|
Editors:
|
Piet Kommers, Tomayess Issa, Theodora Issa, Pedro Isaías and Wendy Hui |
|
Year:
|
2019 |
|
Edition:
|
Single |
|
Keywords:
|
Migration, Prediction, Cloud |
|
Type:
|
Full Paper |
|
First Page:
|
71 |
|
Last Page:
|
78 |
|
Language:
|
English |
|
Cover:
|
|
|
Full Contents:
|
click to dowload 
|
|
Paper Abstract:
|
Virtualized environments are common in cloud services. In a virtualized environment, the efficiency of resources is improved by overcommitting. However, excessive overcommits may overload a virtual machine (VM) and subsequently affect other VMs. Hence, methods that predict the load are used to appropriately place VMs. The load of each VM is typically predicted by using the CPU utilization, but this rate can fluctuate greatly. Consequently, a certain margin is common in the predicted value. However, if VMs have greater variations than expected, overload is possible. If a load is predicted to exceed the capacity, some VMs in the host should migrate to other hosts with more room. The live migration technique is used for migrations while running applications. However, if the margin setting is inaccurate, live migrations will be inefficient. In this paper, we propose a method to perform a live migration using dynamic margin setting. This method gives larger margins to VMs with large fluctuations and smaller margins to VMs with small fluctuations. This method also allows more appropriate predictions to be performed when the fluctuations in the CPU usage is large. Additionally, for stable VMs, resource efficiency may be increased by setting a small margin. In the experiments, this method shows better results than static margin setting, and it is expected to reduce SLA (Service Level Agreement) violations and improve resource utilization efficiency when used for live migrations. |
|
|
|
|
|
|
