Research On SaaS Service Deployment Optimization For Dynamic Resource Demand

SaaS(soft as a service)is a business model that breaks the traditional software sales model.In order to provide a stable and targeted service platform for the development and operation of SaaS services,SaaS service providers build a SaaS platform to manage a large number of SaaS services in a unified way,responsible for the development,deployment,combination and maintenance of software,and obtain stable and sustainable income from their customers.However,how to deploy SaaS to maximize the benefits of SaaS service providers is an important issue for SaaS platform to consider.As the number of users accessing SaaS services on SaaS platform changes,the resource demand of SaaS services changes accordingly.At this time,a single service deployment scheme can not guarantee the stability of the system.Therefore,in the operation process of SaaS platform,it is necessary to constantly adjust the deployment scheme of SaaS services.In the current research,most of them maximize the benefits of SaaS service providers by minimizing the rental cost of virtual machines;for the algorithm in the initial deployment stage of SaaS service,it is easy to fall into local optimization,and the solution quality is not high;for the dynamic deployment stage,most of them adjust the service deployment passively,which is easy to cause the decline of service quality;for the service migration strategy,there is a lack of SaaS service resource demand and virtual machine resource The consideration of correlation,complementarity between consumption and interaction between SaaS services.In view of the above problems,this topic will propose the corresponding solutions.This topic takes maximizing the benefits of SaaS service providers as the research goal of service deployment optimization,and comprehensively considers the impact of virtual machine rental cost and network communication cost generated by interaction and call between SaaS services in the data center on SaaS service deployment optimization.In the initial stage of SaaS service deployment optimization,this paper proposes SaaS service deployment optimization algorithm based on simulated annealing and particle swarm optimization hybrid algorithm,establishes the solution model of the problem and determines the optimization objective.The experimental results show that the algorithm has a good solution effect.In the research of service migration strategy in the dynamic change stage of resource demand,this paper proposes a resource demand prediction model based on Prophet LSTM combination.The experimental results show that the combined resource prediction model has higher prediction accuracy than the existing prediction model.In the dynamic deployment and optimization stage of SaaS services,the service migration timing judgment method proposed in this paper is based on the resource prediction model to better guarantee the service quality of SaaS services;then,aiming at the problem of service screening to be migrated and the problem of virtual machine screening,it analyzes the resource correlation and complementarity between SaaS services and virtual machines,as well as the interaction characteristics between SaaS services,we propose the shortest Euclidean distance moving out SaaS service screening strategy and the shortest Euclidean distance virtual machine screening strategy.Experiments show that the method proposed in this paper can effectively reduce the number of SaaS service migration and virtual machine overload,ensure the stability of virtual machine and SaaS service quality,effectively reduce the network communication cost caused by SaaS service interaction and the resource use cost of SaaS platform,and maximize the benefits of SaaS service providers.