Software Aging in Virtualized Environments: Detection and Prediction
Lists
Only you see this
Abstract
Software aging has been cited in many scenarios including Operating System, Web Servers, Real-time Systems. However, few studies have been conducted in long running virtualized environments where more and more software is being delivered as a service. Furthermore, state-of-the-art methods lack the ability to deal with miscellaneous upper applications and underlying systems transparently in virtualized scenarios. In this paper, we detect aging phenomenon by conducting experiments in physical and virtual machines and identify the differences between the two, and propose a feature code-based methodology for failure prediction through system call, then implement a prototype in virtual machine manager layer to predict failure time and rejuvenate transparently, which is suitable in virtualized scenarios. The evaluation shows the prediction deviation against reality is less than 10%.
BibTeX
@inproceedings{cui2012software,
author = {Lei Cui and Bo Li and Jianxin Li and James Hardy and Lu Liu},
title = {{Software Aging in Virtualized Environments: Detection and Prediction}},
booktitle = {ICPADS},
year = {2012},
url = {https://openreview.net/forum?id=-v5RazOYZ5h},
}Loading related content...