Research on panoramic anomaly detection method for super-scale heterogeneous cloud resources
Junbing Pan, Anni Huang, Xiaoying Mo, Boling Chen, Chunzhi Meng · 2023
The current conventional panoramic anomaly detection methods for heterogeneous clouds mainly achieve anomaly detection by collecting node metrics and comparing them with conventional metric values, which leads to poor detection results due to the lack of effective processing of monitoring sequences. In this regard, a panoramic anomaly detection method for ultra-large-scale heterogeneous cloud resources is proposed. By collecting heterogeneous cloud index parameter samples, constructing dynamic monitoring sequences, and decomposing heterogeneous cloud resource data by combining wavelet transform method to extract heterogeneous cloud index data features, based on which the panoramic anomaly detection method is conceived. In the experiment, the designed panoramic anomaly detection method is tested for its detection effect. The final results can prove that the algorithm has a high AUC value and a more desirable panoramic detection effect when the proposed method is used to detect heterogeneous cloud resources.