Research On Virtual Machine Fault Early Warning In Cloud Computing Environment

With the development of cloud computing and the rapid growth of data size,a single device has been unable to meet the current computing requirements,and in order to improve computing ability and device utilization,virtualization technology is widely used.However,the performance of the virtual machine directly affects the stability and reliability of the cloud platform.On the basis of researching and analyzing the method of fault early warning of virtual machine,this paper deeply studied and researched Extreme Learning Machine(ELM)algorithm and applied it to the fault early warning of virtual machine.Firstly,this paper makes a brief introduction to the ELM algorithm.However,due to the fact that the input weights and offsets of the ELM algorithm are randomly generated,which leads to the instability of the ELM algorithm,the ELM algorithm has its own deficiencies.In this context,by using the global search capability of the fireworks algorithm to optimize the input weights and offsets of the extreme learning machine,this paper proposes an optimization algorithm which combines the fireworks algorithm with the extreme learning machine.Compared with the basic fireworks algorithm,all of the optimized algorithms involved in this paper can improve prediction accuracy and generalization ability of the extreme learning machine in varying degrees.For different improved algorithms,different virtual machine fault early warning models are proposed in this paper.Moreover,the experimental results show that the optimized fault early warning model generates a better prediction accuracy.Secondly,based on the study of fireworks algorithm,this paper proposes a new global optimization algorithm,namely dandelion algorithm,inspired by the dandelion sowing behavior.Compared with the existing algorithms,the dandelion algorithm boasts its great advantages in terms of both convergence speed and accuracy.Therefore,this paper applies the dandelion algorithm to the optimization extreme learning machine so as to further improve its prediction accuracy and generalization performance.Based on this,this paper proposes an optimized model of fault early warning of virtual machine in accordance with the dandelion algorithm and ELM.Experiments show that such model is feasible and practicable.Finally,this paper involves the design and implementation of fault early warning system,and verifies the practicality of the system through tests.The experimental results show that this method can effectively prevent the anomalous problems and ensure the sound operation of the service.