Performance Optimization At Software Architecture Level Based On Surrogate Model

Architecture-based software performance optimization can help to find performance problems and mitigate their negative effects at the early stage.According to traditional application,some rule-based and metaheuristic-based performance optimization methods at software architecture(SA)level have been proposed in recent years.However,in the most existing rule-based methods,the count and the order and improvement amplitude of each rule have not been fully considered in search process.So the search space is limited and better solutions may be excluded.Most metaheuristic-based methods cannot obtain high solution quality and need long optimization time due to the mixed coding and performance evaluation process.In recent years,with the rapid development of cloud computing technology,people's vision of SA layer performance optimization has gradually expanded to the cloud application.In view of the SA layer performance optimization of cloud applications,people have proposed a two stage optimization method.In the process of optimization,the solution of response time constraints is a high computing cost operation,which makes the whole optimization process more time-consuming.To address these problems,this paper studies the performance optimization of SA layer based on surrogate model.Our research contents are as follows.(1)A methond for Performance Optimization at Software Architecture Level based on Surrogate Model for Traditional Application(TSAPO)is proposed.This method first constructs an optimization model MOM4SAO which can effectively increase the search space,and designs a differential evolution optimization algorithm(DESM4PO)for solving the model.The algorithm uses a variety of crossover and mutation strategies,and uses the surrogate model to estimate the response time value of the system to improve the quality of the solution and reduce the time-consuming.(2)A methond for Performance Optimization at Software Architecture Level based on Surrogate Model for Cloud Application(CSAPO)is presented.This method first defines a cloud application SA layer performance optimization model named CAPOM.This model can increase search space and provide support for searching the best cost performance of cloud resources.Therefore,a cloud performance SA optimization algorithm MODE4CA is proposed to solve CAPOM model efficiently.(3)A framework for the optimization of SA layer performance evolution is designed.(4)The TSAPO is compared with the metaheuristic-based PCM method in 4 different scale cases.The experimental results show that TSAPO in the three indicators of contribution,generational distance and hypervolume is significantly better than PCM,and effectively reduces the time-consuming.The experimental research are carried out between CSAPO and two stage Ardagna in two different scale cases.The experimental results show that CSAPO compared with Ardagna can obtain lower cost of resource use in cloud resources scheme.