Research And Implementation Of Node Anomaly Detection Method In Cloud Computing Environment

In recent years,the rapid development of cloud computing technology has been widely used in many fields.The cloud computing environment aggregates a large number of physical and virtual resources and provides multiple levels of services.Various services ar e deployed on nodes in different ways under the cloud computing environment.Due to the diversity of applications and the dynamic deployment environment,cloud computing environment often have anomalies,which can be transformed into critical failures,services unavailability and resources waste and bringing severe challenges to the reliability of cloud computing.In order to maintain the high reliability of cloud computing,more and more experts and scholars focused their research on anomaly detection and proposed node anomaly detection methods under cloud computing environments.However,the existing methods have the following problems:(1)For the method of detecting the status of a single node based on the node's running data,data changes caused by norm al system events,resource calls,and other behaviors within the node may be detected as abnormal behaviors,which will eventually lead to a high false positive rate of the abnormal detection method.(2)Due to data transmission and resource sharing among nodes,an abnormality in the node may cause anomalies in other nodes.The existing research does not consider the problem that the node exception propagation leads to the associated node anomaliesThis paper focuses on the problems of cloud computing node anomaly detection methods and combines the characteristics of cloud computing.Node anomaly detection methods are studied to improve the reliability of cloud computing.Node anomaly detection method in cloud computing environment is proposed.The node anom alies in the cloud computing environment are analyzed and detected.The main tasks include.(1)A node anomaly detection structure based on Agent-based under cloud computing environment is proposed.A node monitoring model is established through this struc ture.The node monitoring model model includes node operation data collection,node relationship graph establishment,and node data transmission.(2)A single node anomaly detection method is proposed.First,anomaly detection methods based on time points are used to detect node data.Then,a time window-based method is used to determine node status when anomalies occur.(3)Associated node anomaly detection method is proposed.Other related nodes are found through the node relationship graph in the case of single node abnormality.Different detection algorithms are used according to the different relationship of the nodes,including direct associated node abnormal detection and indirect associated anomaly detection method.(4)A prototype of node anomaly detection software under cloud computing environment is designed and implemented,and the effectiveness of the method is verified through related experiments.The experimental results show that the single-node method proposed in this paper can filter the influence of noise,and has lower false positive rate and higher accuracy.The associated node algorithm can find the associated node anomaly and has a high accuracy in the case of single-node anomaly.