Anomaly Detection for Container Cluster based on JointCloud Platform
Zhengmin Li, Zhaoxin Zhang, Xinran Liu, Chunge Zhu · 2019
In order to accurately discover container exception data of large-scale container clusters to guide the maintenance of container clusters, a new anomaly detection model for container clusters is proposed in this article. The model combines the advantages of supervised learning and unsupervised learning to accurately and efficiently label container anomaly data in a large-scale data environment. Experiments show that the labeling rate of the raw data is as high as 95.6%, and the accuracy of anomaly detection is as high as 87.0%. Simultaneously, the common five classification algorithms are used to compare the anomaly detection effect between the labeled data and the raw data, and the validity of the model is further verified.