Research And Application Of Virtual Machine Anomaly Detection Technology In Cloud Platform

Nowadays,cloud computing is widely studied and applied in global IT companies.The development of is increasingly inseparable from cloud computing.The most critical breakthrough in cloud computing is the change in the way resources are used,which can be virtualized to quickly create a stand-alone,on-demand virtual machine for users to use.Real-time performance monitoring and accurate status detection of virtual machines makes it possible to prevent and deal with the virtual machine before a major failure occurs.This is an important issue that needs to be studied at present.In order to solve this key problem,this thesis proposes an anomaly detection strategy based on the operational environment for detection domain partitioning and a Canopy-Kmeans anomaly detection algorithm.And this thesis designs a framework of the virtual machine anomaly detection system in cloud Platform.The main research work and innovations of this thesis are as follows:(1)This thesis studies the infrastructure of cloud platforms in depth,especially the infrastructure-oriented.And aiming at various characteristics of the current cloud platform,this thesis proposes an anomaly detection strategy based on the operational environment for detection domain partitioning.According to the strategy,this thesis designs a framework of the virtual machine anomaly detection system in cloud Platform.These provides a reliable guarantee for the stable operation of the cloud platform.(2)The thesis deeply studies the system structure of OpenStack that is an open-source cloud platform,and uses it to build a cloud platform experimental environment.Taking Open Stack as the research object,according to its characteristics,this thesis proposes the virtual machine running Status Information Index system,which is divided into two kinds: virtual machine running Environment Index and virtual machine performance index.Taking Virtual Library of Open Stack as the research object of virtualization,this thesis designs and implements the data collection module of the virtual machine anomaly detection system in the cloud platform.According to the characteristics of cloud platform and collection module,this thesis proposes an adaptive collection strategy.(3)By studying anomaly detection techniques,which are mainly anomaly types and data anomaly analysis methods,this thesis proposes an anomaly detection method for virtual machines in cloud platforms.Firstly,we classify all the virtual machines in cloud platform by using the detection domain partitioning strategy.This strategy uses the canopy algorithm to make rough clustering of the data of the running environment indicator set.Then,the PCA algorithm is used to reduce the dimension of the data of performance indicator set in the detection domain.Finally,the Canopy-Kmeans algorithm is used to detect abnormal data after dimensionality reduction.(4)This thesis simulates the real cloud platform environment by means of anomaly injection,and the experimental results verify the accuracy of the anomaly detection of the virtual machine anomaly detection system.