Research On SaaS Service Deployment Optimization For Dynamic Resource Demand

Saa S(Soft as a Service)is a new service-oriented software delivery model based on the Internet.Tenants can access the leased Saa S service over the network without purchasing a complete software service and configuring the appropriate hardware system.In order to manage a large number of Saa S services,Saa S service provider deploy Saa S services on the Saa S platform,responsible for software upgrade and hardware maintenance.How to reasonably deploy Saa S services on the corresponding virtual machines makes Saa S services high-performance and low-cost operating,which is an important issue to be considered by the Saa S platform.There are a large number of Saa S services on the Saa S platform.There are complex combinations between Saa S services and virtual machines.The dynamic resource demand of Saa S services make the effective deployment of Saa S services more challenging.At present,the research on the initial deployment of Saa S services is to convert the problem into a combinatorial optimization problem,but there is a problem that the quality of the solution is not good and it is difficult to search for the optimal solution;in the dynamic deployment process,most of the research adopt the passive response deployment adjustment strategy and The relevance of the dynamic resource demand of Saa S services is not taken into account.This paper studies the deployment optimization of Saa S services for dynamic resource demand.For Saa S services initial deployment optimization problem,this paper establishes the mathematical model to solve the problem and proposes a Saa S service deployment optimization algorithm based on discrete beetle swarm optimization,studies how to search for better solution according to the characteristics of Saa S service deployment problem.The solution proves that the algorithm has better optimization ability for Saa S service deployment.For Saa S services dynamic deployment optimization problem,this paper adopts Prophet time series model to predict resource demand,and then proposes a migration timing judgment method based on resource demand prediction.By predicting resource demand trend based on historical resource demand,the paper makes active deployment adjustment to better guarantee service quality.For the selection of Saa S services moving out and the selection of virtual machines moving in when Saa S services are migrated during the dynamic deployment process of Saa S service,the selection strategy of moving out Saa S service with the maximum resource relevance and moving in virtual machine with the minimum resource relevance were designed through the quantitative analysis of the dynamic correlation between Saa S service resource demand and virtual machine resource demand,and the experiment verified that the resource-related Saa S service migration strategy proposed in this paper could effectively guarantee the service quality,reduce the rental cost of virtual machine and reduce the migration times of Saa S service.Based on the initial deployment and dynamic deployment algorithm,this paper designs and implements a Saa S service deployment management prototype system,The verification result of the prototype system shows that the optimization algorithm proposed in this paper can implement the effective deployment of Saa S services,improve the ability of Saa S platform to manage the deployment of Saa S services,and provide an effective solution for the high-performance and low-cost operating of Saa S services.