摘要
基于隐马尔可夫模型(Hidden markov model,简称HMM)的虚拟机性能,提出了一种虚拟机性能异常的预测方法。该方法的核心思想是基于业务系统的运行时资源消耗具有一定的规律性。通过该规律性采用隐马尔可夫模型刻画当前业务系统的正确状态,并根据业务系统预测结果是否偏移正常状态来判定业务系统是否出现性能异常。基于TPC-W的试验结果显示,该方法具有快速发现和定位性能异常的能力,且其运行时开销较小。
We made an unusual discovery of virtual machine performance based on Hidden Markov Model. The core idea is that the consumption of resources has a certain regularity when the operation system runs. Hidden Markov model is adopted to correctly portray the current state of business systems, and according to the results of forecasts to determine whether the business system is abnormal or not. The test results based on TPC-W show that this method has the unusual ability to quickly find and locate properties, and the cost is much lower.
出处
《河南农业大学学报》
CAS
CSCD
北大核心
2016年第4期563-567,共5页
Journal of Henan Agricultural University
基金
河南省教育厅科学技术研究重点项目(14A520084)
河南省科技厅科技攻关课题(152102310325
152102310118)
关键词
虚拟机
性能异常发现
机器学习
virtual machine
discovery of property abnormality
robotic learning
