Research On Fault Detection Method In Cloud Computing Environment Based On Uncertainty Reasoning Algorithm

In recent years,with the continuous development of cloud computing technology and the continuous expansion of the application field,cloud computing technology has become one of the hottest technologies in the IT industry.Due to the complexity,dynamics,resource sharing,large scale and other characteristics of the cloud computing environment,the cloud computing environment is more and more prone to fault,or even the failure of cloud computing services,which brings severe challenges to the availability and stability of cloud computing.In order to ensure that the cloud computing environment can provide continuous service and to ensure the high availability and stability of the cloud computing environment,more and more experts and scholars at home and abroad on the cloud computing environment of fault detection is studied,and put forward many cloud computing fault detection methods and detection framework,realize a lot of the detection system of the relevant methods,but the characteristic data or performance data in cloud computing environment system based on the fault detection method has some problems: 1.For the fault detection method using fault model,it is not easy to conduct fault detection through the fault model when there is an unknown fault type in the cloud computing environment,affecting the accuracy of fault detection;2.When unknown faults occur in the cloud computing environment,it is difficult to continuously update the fault model.Based on the structural characteristics of the cloud computing environment,this dissertation aims to solve the problems existing in the fault detection of the cloud computing environment.The main contents of this dissertation are as follows:1.Fault model building and known fault library building,cloud computing environment fault model will be built according to cloud computing environment architecture,fault injection method is adopted to acquire fault data and build known fault fault library;2.Known fault detection: after the dimensionality reduction of data is realized through the selection of performance characteristics,the known fault detection is carried out using the gaussian naive bayesian algorithm.After the known fault detection,there are still abnormal data that cannot be detected,and then the unknown fault detection is carried out.3.Unknown fault detection,first of all,using cloud computing environment runtime data calculation coefficient of multiple linear regression construct multiple linear regression model,using multiple linear regression model realizes fault detection of abnormal data,so as to realize the unknown fault detection,and then the method based on density clustering generate new fault type,fault data aggregation by artificial add tags,realize fault library update.4.Designed and implemented a cloud computing environment fault detection system prototype,conducted fault detection experiments by building a real cloud computing environment,and finally confirmed the effectiveness and feasibility of the fault detection method proposed in this dissertation.