Research And Implementation Of High Concurrent High Available Improvement Of Online Shoe Shopping System

The shoe shopping online system is a B2C-based online footwear trading website.In recent years,with the popularity of Jingdong Mall,Taobao,and other large electronic websites,the awareness of users purchasing goods online has grown rapidly.With the increase in the number of users,single service and high coupling system architecturet and technology of the old system can not handle the large number of concurrent requests,which will result in long response time,poor user experience,and even system collapse in extreme cases.The normal operation of the website caused a lot of economic losses to the company.This topic mainly analyzes the causes of system concurrency performance limit from server hardware,software,system architecture,application technology,etc.,and looks for solutions to improve concurrent performance.At the database level,the table structure design of the original single-library single-table is divided into 8 libraries,256 tables per library,and the read and write separation is realized by Amoeba to finally improve the SQL query efficiency of the single-library single-table.In terms of distributed caching,by using Redis,the direct interaction between user requests and the database is reduced,thereby optimizing the efficiency of data reading.In terms of search engines,comparing the advantages and disadvantages of today's two mainstream search engine products,Solr is selected as the full-text search instead of the direct query database in the product information retrieval to improve the retrieval efficiency.In order to improve the ability to process data,increase the throughput of the system,use Nginx as load balancing,and use the minimum connection scheduling algorithm to dispatch user requests to the most suitable server to improve the concurrent performance of the system.On the other hand,when the product is put on the shelf,the template fill data is generated by the Freemarker technology to statically make the detail page of the product to increase the running speed.Finally,by splitting the original single service architecture into multiple microservices,the services are invoked asynchronously and the epoll I/O model and Reactor design pattern are used to ensure the high efficiency of the service to improve the concurrent performance of the system.Through the above method,after implementing the improvement of the shoe shopping online system,the query response time of the order form table is reduced from the original average 6000 ms to an average of 2000 ms.Tens of thousands of products and product retrieval speeds dropped from 6.9s to 0.003 s,and service response time dropped from an average of 107 ms to an average of 18 ms.The results show that this research and system improvement have greatly improved the concurrent performance of the shoe shopping online system.