Research And Design Of High Performance Web Service Based On REST Style

In age of big data, network applications are generally faced with high concurrency problems, and this kind of problem has very great importance. In general condition, when application systems encounter high concurrency problem, they will take corresponding measures according to their circumstance to deal with it. For traditional large web sites, the measure is usually to increase hardware indicators, such as adding servers, to solve the concurrent pressure caused by too much users. But the way above can not fundamentally solve high concurrency problem.In this paper, we study how to design a high performance Web services application system based on the design style of REST service, from the existing challenges facing highly concurrent application system. Our work is about how to improve the concurrent performance of the system in the case of high concurrency. The emphasis of this research is the efficient use of Node.js and key technologies of improving the ability of concurrent application, such as load balancing algorithm.We make a research on current technology platform of Node.js. Facing the problem of multi-core CPU running on single-threaded Node, the existing solutions have the disadvantages of great utilization of system resources and uneven utilization of multi-core resources. We first make a comparative study of the existing programs, and then distribute requests simply and efficiently using Node itself instead of the reverse proxy server, improving the previous problem of uneven multi-core CPU load scenario.We also make a research on the existing load balancing algorithm, detailed analyze of WRR equalization algorithm and making an optimization and improvement of it through Nginx. We feedback the real-time response status of back-end server to the calculation of the weight of the front-end load, and we give different weights to different requests when calculating the average length in response, thus making up inflexible defects of the original algorithm. The experiments verify that the amount of concurrent peak of environmental sustainability upgrade from 350 to 450, making a good improvement comparing to the original algorithm.