Design And Implementation Of Intelligent Optimization Platform For Software System Parameters

With the rapid development of the Internet and computer science and technology,the use of software systems to solve complex practical problems has become more and more abundant.However,with the continuous increase of business access,the load of the software system is increasing and the performance is decreasing.It is urgent to improve the performance of the software system to adapt to more application scenarios.The performance of a software system largely depends on its own parameter configuration.Many organizations rely on hiring experts to configure parameters,which is often very expensive.And with the increasing scale and complexity of software systems,the traditional manual optimization of parameters is increasingly difficult.How to achieve automatic optimization of the parameter configuration of the software system has become a research hotspot.Regarding the issues above and commissioned by the company,this thesis designs and implements an intelligent optimization platform for software system parameters.The system is mainly based on the current mainstream and latest parameter optimization algorithms in academia and industry.It supports users to configure the optimization experimental environment and customize the combined parameter optimization algorithms according to the performance requirements of the system to be optimized in the actual business scenarios,to build an optimization experiment,and finally obtains the parameter optimization configuration of the system to be optimized,thus effectively improves the performance of the software system and reduces the operating cost.The main work of the thesis are as follows:(1)Requirement analysis.Based on the research and analysis of the existing parameter optimization work of software systems,the thesis confirms the requirements of the intelligent optimization platform for software system parameters,and refines the business into a user management module,a system management module to be optimized,an algorithm management module and an experiment management module according to the optimization workflow of the parameter configuration of the target software system,and explains the performance requirements of the system;(2)Design and implementation.The thesis designs the overall architecture according to the requirements,completes the development of the function modules and optimizes the parameters of software system intelligently.Based on the Spring Cloud microservice architecture,it builds the business logic of each functional module to realize loosely coupled and highly scalable functional modules of the system.Base on the improved token bucket algorithm,the service restriction strategy of microservice is designed to improve the service fault tolerance of the system.Based on Open Stack virtualization tools,the creation and management of the optimized experimental environment is completed,and the automatic installation and deployment of the environment is realized.Based on Docker and Kubernetes technology,the life cycle management of algorithm components is realized to ensure the independence and stability of algorithm operation;(3)System test.After designing and implementing the system,this thesis composes test cases based on the requirement analysis and builds a test environment for system testing.The system test part chiefly tests the main functions of the four functional modules of the platform and tests whether the response time of each module meets the performance requirements,whether the system interaction is stable,and whether the algorithm used has improved the fault tolerance of microservices.After system testing,all the functional logic processing of the system designed and realized is normal,and the response time meets user expectations,and the system has sufficient stability and reliability.After the platform goes online,users can automatically generate an optimized experimental environment through simple configuration,and finally,get the optimal configuration results of the target parameters of the software system to be optimized under the given hardware environment.By avoiding traditional manual parameter optimization,the cost of parameter optimization can be effectively reduced,thereby realizing the intelligent demand for parameter optimization of the target software systems.To sum up,the platform can improve the performance of software system as much as possible based on no increase in operating costs,thus reducing the operational cost of the software system,it is of great significance in promoting the automatic optimization of the performance of software systems.