Research On Strategy Of High Concurrency Web Pages

With the development of the network, surfing the Internet become a lot easier and it leads to the increasing access amount of the websites. In nowadays, the experience of using the Internet products is concerned by every website's owners. Make sure every user can view the pages and the website is stable under high concurrency is an important problem to solve. By using monitor programs, we found out that the main reason of non-responding pages is fail to get the data from data servers. And no data transfer to the application server means the page cannot be build and response the user's requests.To solve this problem, we propose two solutions: use a distributed caching system or static pages. Build a distributed cache system between the data server and application server. When application needs to get data from data server, it request data from cache server first. The data of cache server is from each time when an application gets data from data server. According this way, we add an extension of data server, and improve the performance of data server by reduce the amount of requests. Static pages is let the application server build the pages that user will view before the viewing requests really come. When user needs to view a page, the application server just transfer the page which is already be build and just stored in the application server. In the case, user's requests will not reach the data server, so no pressure will be add to the data server.This paper test the performance of three kind of pages: pages without using any additional processing, pages using cache to accelerate, static pages. By compare the result of these three kinds of pages with different testers. We can find out that using the caching system can make the pages response faster and let the whole system be stable. And static pages will response a lot faster than using cache. As the result, we can use caching system to accelerate the whole website and static the pages which are viewed most. In this way, we can get a good responding rate at a lower price.