An Adaptive Resource Allocation Method For Web Applications In Cloud Computing

The adaptive allocation problem of cloud resources for Web application is a kind of im-portant cloud resource management problem,Widely existing in cloud service providers.With the increasing scale of the cloud,the effective allocation of cloud resources will significantly increase the economic benefits of cloud service providers.And where energy consumption is increasingly becoming an important issue for cloud service providers' costs.In addition,the VM provision and VM placement problem of Adaptive resource allocation for cloud resources is the NP-hard problem.Therefore,with the minimum energy consumption as the optimiza-tion target,how to integrated the whole cloud resource adaptive configuration mechanism and find the effective cloud resource adaptive configuration algorithm has significant theoretical and practical value.In this thesis,the adaptive allocation problem of cloud resources for Web application is described and analyzed.Then we construct mathematical model of it.Finally we propose the corresponding algorithm in three stages.In the first stage,we based on fuzzy clustering,propose the improved BP neural network algorithm to predict the request load.In the second stage,the SAQ-learning algorithm is proposed by using the request load from the first stage and the amount of virtual machines to get the optimal virtual machine provision strategy.In the third stage,the method of measuring the cosine similarity of the remaining space resources is proposed according to the proportion of the resources of the remaining space,and a heuristic method,the best balance fit method is proposed based on the best fit algorithm.The minimum energy consumption is obtained from the perspective of cloud service provider.In order to verify the efficiency and effectiveness of the proposed algorithms.We use the simulation experiment and the ANOVA technique to select the parameters in the algorithm and analyze the algorithm module.The proposed algorithm is compared with the general algorith-m which solves the similar problem at present by generate different scale test data sets.The experimental results show that the proposed algorithm outperforms the existing algorithm.