Unsupervised anomaly prediction method for two-stage cloud server

Abstract

The invention discloses an unsupervised anomaly prediction method for two-stage cloud server, which is used for solving the anomaly prediction problem of a cloud server environment. The method comprises a prediction stage and an anomaly detection stage, the prediction stage is used for training a many-to-many time sequence prediction model according to preprocessed historical cloud server key performance index data, and the model is used for predicting cloud server KPIs data at the future moment; in the anomaly detection stage, a multivariable anomaly detection model is trained according to preprocessed historical cloud server KPIs data, anomaly detection is conducted on the predicted KPIs data at the future moment through the model, the anomaly probability of data points at the future moment is obtained, finally, an anomaly probability threshold value is set, and the data points with the anomaly probability threshold value larger than the threshold value are regarded as abnormal datapoints. Otherwise, the data points are normal data points, and an anomaly prediction result is obtained. The invention has the advantages of being independent of label data, wider in applicability andexcellent in performance.

Description

technical field [0001] The invention belongs to the field of computer application technology, and in particular relates to a two-stage cloud server unsupervised abnormal prediction method. Background technique [0002] Cloud server (Elastic Compute Service, ECS) is a computing service that is simple, efficient, safe and reliable, and has elastically scalable processing capabilities. It makes developers no longer need to purchase hardware separately, and only needs to host the service on the cloud server. Although cloud server hosting services have very broad prospects, cloud server providers still face many challenges, among which the availability of cloud servers ranks at the forefront. Software and hardware failures or operational errors may directly lead to cloud server downtime, resulting in losses to users. Therefore, how to effectively monitor cloud servers and make effective anomaly predictions has become an important research issue. [0003] The monitoring data of ...

AI Technical Summary

Problems solved by technology

[0004] At present, researchers have proposed many anomaly prediction methods. The document "Anomaly Prediction Method, Abnormal Prediction System and Abnormal Prediction Device (CN106330852B)" proposes to record and construct corresponding prediction rules according to the system command to perform abnormal prediction. This method is not applicable to the system In complex cloud environment scenarios, there are many commands in the system layer and application layer of the cloud environment scenario, and it is difficult to predict the abnormality of the cloud server from this layer

The document "A Method and Equipment for Abnormality Prediction of Drive System (CN110426634A)" proposes to construct a state distribution map of the system based on the current response parameters and historical state of the system, and quantify the abnormal risk in the system area, so as to predict the abnormality of the system. However, There are complex dependencies between cloud server KPIs data, and they are unbalanced and unlabeled. This method is particularly difficult when constructing a state distribution map, which makes it difficult to build an anomaly prediction model.

Therefore, how to construct a method suitable for cloud server anomaly prediction is still a challenge for cloud server intelligent operation and maintenance.

Images