Design And Implementation Of High Concurrency Shopping Platform System

In recent years,with the rapid development of the Internet,the number of users of e-commerce websites has continued to increase,and the advantages of e-commerce traffic and its convenient and efficient features have completely changed the traditional shopping methods.However,as the concurrency of the system becomes higher and higher,the peak period of concurrency will cause the system to respond too long,and even cause the system to crash due to the inability to load high concurrent traffic.This undoubtedly has a significant impact on the traditional monolithic Web system architecture.Here comes a new challenge.Therefore,a mall shopping platform that can provide high-efficiency,high-concurrency and high-availability services has important practical significance and application value.Based on the actual needs of the company's project and combined with specific business scenarios,this paper designs and implements a distributed shopping mall system that supports high concurrency.The main research work of this paper includes:(1)Adopting the distributed service architecture Dubbo as the basic architecture of the platform in the system architecture,providing the system with high-performance service calling schemes,and splitting the system architecture level vertically;using SpringBoot at the application level to split the application services horizontally based on the concept of microservices The application services are deployed and run independently;Zookeeper is used as the service monitoring center to complete the management and scheduling between the server and the consumer,it provides isolation between services and improves system scalability.(2)Optimize the design of the database structure,and use Redis as the distributed cache of the database,which can greatly improve the storage and access capabilities of massive data;use read-write separation to realize read-write requests sent to different servers for processing,and improve the system's processing request Efficiency,the configuration of the master-slave replication scheme effectively guarantees the safe backup of data and avoids data loss due to server downtime;using MyCat to split the database table structure horizontally and vertically will help speed up the response time of data acquisition.(3)Realize single sign-on based on JWT and realize cross-language support login authentication.Compared with traditional Cookie and Session login schemes,the information security is higher,cross-domain resource sharing can be realized,and the repeated login of different subsystems is avoided at the same time.Situation:Realize distributed locks based on Redis and Lua,which solves the safe operation of shared data in multi-threaded scenarios.Compared with traditional distributed locks,it has higher data security and better performance.(4)In order to avoid server downtime caused by excessive instantaneous traffic,a current limiting strategy is designed based on the leaky bucket algorithm and the token bucket algorithm,and the appropriate algorithm model is selected according to the different characteristics of the two algorithms to process high concurrent traffic requests according to the business scenario.,Which guarantees the sustainable provision and stability of system services.(5)In order to increase the concurrency of the system,a dynamic load balancing strategy is designed.Based on the weighted round-robin load balancing algorithm,the weight calculation is optimized,using the respective of CPU,disk I/O,memory and network bandwidth.The utilization rate and other influencing factors calculate the comprehensive weights,request allocation is more reasonable,and the cluster status is monitored in real time,and reasonable adjustments are made to improve the request processing connection success rate and effectively improve the system load capacity and concurrency performance.Based on the above architecture design,a distributed shopping integrating portal display function module,back-end management function module,search function module,product display function module,shopping cart function module,order function module,and "snap buying" activity function module has been developed.Mall system.Finally,a detailed and specific test and analysis of the platform's functions and performance are carried out.The test results are in line with the estimated results,and each functional module is currently operating normally.Using Jmeter to stress test the high concurrency situation,comparison of results with traditional weighted round-robin load balancing algorithm,the concurrency of the dynamic load balancing strategy adopted in this system has increased by about 11.2%,and the request response time has been reduced by 5.4%.More suitable for high concurrency environment.The platform designed and developed in this paper has been put into operation.The system runs stably and smoothly.It can provide users with efficient and stable services and enhance the user's shopping experience,indicating that the system has certain application value in solving the problem of high concurrent access.