摘要
针对分布式平台资源利用的特征,在集群局部资源利用密度异常情况检测技术的基础上,引入时间序列的检测方法,提出了一种对集群资源利用率呈现缓慢增长的异常情况进行检测的算法。最后通过某高校基于Hadoop服务器集群采集到的数据进行异常检测,验证了提出的基于最近邻结点资源异常检测方法的准确率和可行性。
The characteristics of distributed platform resource utilization are introduced. Based on the cluster local resource utilization density anomaly detection technology, the time series detection method is brought to detect the abnormal situation of slow growth of the cluster resource utilization rate. Finally, a anomaly detection of the data of a certain university based on Hadoop server cluster is used to verify the accuracy and feasibility of the anomaly detection method proposed.
作者
杨安
孙彦超
YANG An;SUN Yanchao(Information and Network Management Center, Beijing Information Science ＆ Technology University, Beijing 100192,China;Office of Educational Administration, Beijing Information Science ＆ Technology University, Beijing 100192, China)
出处
《北京信息科技大学学报（自然科学版）》
2017年第6期58-62,共5页
Journal of Beijing Information Science and Technology University
基金
北京市教委科研计划资助项目(KM201711232023)
