Research On Software Aging Modeling Method Of Openstack Cloud Platform

Recent years,cloud-based services have been used in many different fields,including health-care,public transportation,mobile communications,etc,and have gradually become the mainstream trend of the Internet industry development.However,the long-time uninterrupted running of the cloud platform system will occur a phenomenon called "software aging".This result to short-term system downtime or instability,which will seriously affect the normal operations of Internet companies.Against the the above software aging problem of cloud platform system,this paper uses two different models to predict and maintain the software aging of the OpenStack platform.The prediction model analyzes the change trend of key performance parameters of the system to obtain the change trend of the system performance parameters in the future.The maintenance model predicts the future state of the system by simulating the changing process of the running state of the system.Through these two schemes,the aging of the cloud platform system is effectively identified,and the time threshold of the appearance of the aging feature is obtained.Within the time threshold of aging occurrence,proactive implement maintenance measures to mitigate the system's performance degradation due to aging,avoid system failure events,and enhance system reliability.This article is based on real project,and the research content includes the following three items:1)Establishing the software aging prediction model of the OpenStack cloud platform system based on HMM.The initial values of the model parameters are set under the common constraint rules of the hidden Markov model.The whole forecasting procedure is divided into two parts:a)Model establishment:Baum-Welch algorithm is used for model parameter training to establish a model,which can truly simulate the system performance changes.Simultaneously,the Forward-Backward algorithm is used to evaluate the training accuracy of the model.b)Model prediction:Based on the HMM model and the observation sequence,the Viterbi algorithm is used to predict the change trend of performance parameters over a period of time in the future,and determine the optimal time point of the regeneration executing,to achieve the purpose of maintaining system reliability.Finally,MAPE and MRSE are used to evaluate the model prediction effectiveness to verify the availability of the prediction model.2)Designing and implementing the aging maintenance model of the OpenStack cloud platform system based on the DFT model.Using dynamic logic gates to describe the non-stationary dynamic evolution process of the system,a solution combining virtualization technology and fault maintenance is proposed,establish an OpenStack cloud platform system with fault tolerance,and construct a dynamic fault tree model of the system.By analy si sing the reliability of the model,the timely deployment of the aging maintenance plan is achieved.After comparing this model with the system model without maintenance,it fully verifies the advantages of DFT in the field of aging maintenance3)Design and plan the construction plan of the OpenStack cloud platform on the physical server system to realize the deployment of the cloud computing platform based on OpenStack.Secondly,in order to speed up the occurrence of system software aging,a load generation scheme for CPU,memory,network and other components is proposed.