Anomaly Judgment for Nodes Based on Agent under Cloud Environment

Under cloud computing environment, cloud computing systems often have abnormalities due to the diversity of applications and the dynamic deployment environment. Collecting node performance data is very valuable for anomaly detection. There are many nodes under cloud computing environment. Many related studies only make abnormal judgments based on the status of single node. However, the state of multi-nodes is not judged. In this paper, a method of nodes anomaly judgment (NAJ method) based on agent under cloud environment is proposed. The collected data are analyzed by using agent technology. Firstly, data conversion technology is used for constructing a unified format for data analysis. Secondly, the detection model is trained and the degree of data anomalies is analyzed through unsupervised learning method. Finally, differences between anomalous nodes and other nodes are detected. The experimental results show that the method can judge the anomaly nodes effectively.