Análise de desempenho de web servers utilizando diferentes backends da plataforma docker

Abstract

In the context of the growing demand for modern web applications, the research presented in this work arises in response to the continuous development of technologies aimed at improving the performance and efficiency of web servers. Web servers, essential for running and storing applications accessed over networks, play a vital role in the user experience. In this scenario, the choice and optimization of these servers become crucial to guarantee fast and effective responses. The advent of containerization, exemplified by Docker, has brought an innovative approach to application deployment, enabling isolated execution and efficient scalability. This study focuses on the performance analysis of web servers in Docker environments, recognizing the relevance of selecting ideal configurations to obtain maximum performance and a good end-user experience. To carry out this work, six scenarios were established that incorporate different backends (runC, gVisor and the WSL component). In addition to the variation in backends, two distinct web servers were outlined, Apache and Nginx. Additionally, it was decided to use two different operating systems, Linux and Windows. As the scenarios were implemented, the benchmarking ApacheBench tool was used. This tool made it possible to collect data on crucial metrics for analysis, including failed requests, requests per second, and time per requests. Statistical tests, such as Shapiro-Wilk and Bootstrap, were employed to substantiate and validate the conclusions derived from the performance analysis. After comparative analysis, it was found that the scenario that presented the best performance was the first scenario evaluated. In this scenario, we used web server Apache, hosted on Docker with backend container and runC, implemented on the Linux operating system. This finding offers valuable answers for Information Technology professionals who play crucial roles in the design and development of web applications, providing them with information that will enable them to offer even better experiences for users.