Space4cloud-based Performance Multi-objective Of Cloud Application Evolutionary Optimization Approach At Software Architecture Level

The performance optimization of software architecture of cloud application(SAoCA)can ensure the performance requirements in the early stage of software development and provide effective support to reduce the cost of future cloud resources.Now the existing performance optimization methods at SAoCA level are often based on foreign cloud service resources and the performance optimization of SAoCA layer is often abstracted into a single objective optimization problem with constraints.These methods can only search limited spaces,and it is difficult to obtain the optimal cost-effective SAoCA.In view of the above problems,this paper studies and proposes a method for evolutionary optimizing the performance multi-objective of cloud application at software architecture layer(POM4CA),based on Space4cloud framework of modeling and performance evaluation at SAoCA level.The main works of this paper are as follows:(1)The cost library about service resource of domestic mainstream cloud is built,based on the domestic mainstream cloud service provider.We conduct a systematic investigation of various cloud services resources provided by Ali cloud,Baidu cloud and HUAWEI cloud and give a pricing model for all kinds of cloud services resources.Combined with the domestic mainstream cloud services resource,this paper constructs the library of domestic mainstream cloud service resource,based on the Space4cloud framework.We can construct 60 different types of domestic public virtual machines using the resource library.(2)A multi-objective optimization model of performance is designed at SAoCA level.In order to accurately describe the relationship between the types and numbers of virtual machine and response time and resource use costs of cloud application at SAoCA level,this paper defines a multi-objective optimization model for performance at SAoCA level based on the library of cloud service resource.Compared with existing methods,the multi-objective optimization model increases the search space at SAoCA level and supports for improvement programs of better quality.(3)A multi-objective differential evolutionary optimization algorithm named MODE4CA is proposed at SAoCA level.Based on the multi-objective optimization model of performance at SAoCA level,combined with the characteristics of the multi-objective differential evolutionary algorithm(MODE),this paper proposes a multi-objective differential evolutionary optimization algorithm at SAoCA level.MODE4CA algorithm can search a group of cost-effective SAoCA.Individual coding,mutation operator with mutation policy pool,crossover operator with crossover policy pool,fitness function are designed in MODE4CA.(4)Finally,in order to demonstrate the effectiveness of the proposed method(POM4CA),a case study is carried out.Two cases of Meeting in the Cloud(MiC)applications and Open For Business(OfBiz)application are used to compare the cost and response time of the POM4CA method and the method of Ardagna.The experimental results show that,the POM4CA method can obtain better response time with lower cost of cloud resource usage,compared with Ardagna method.In the worst case,the SAoCA solutions obtained by the POM4CA method also has 13.2%solutions in terms of resource usage cost lower than the method of Ardagna,and 55.8%of these solutions have better response time than the method of Ardagna.In this paper,the POM4CA method can be used to develop cost-effective software architecture of cloud applications which are deployed on the domestic mainstream cloud service providers.It has certain academic significance and good application value.