Web Hosting Performance Measurement. Approach of a Web Hosting Benchmark in Times of Virtualization

Resumen o Introducción

In this thesis, an approach for evaluating the performance of web hosting solutions is developed to facilitate the comparison of offerings from different providers and aid companies in choosing a web hosting service. The focus of the research is on identifying and investigating methodologies for collecting reproducible performance metrics within the constraints of common virtualized web hosting environments, using only PHP and MySQL functions. A comprehensive review of existing literature and research in web development and web hosting provides the theoretical basis for benchmarking web servers and measuring web performance. Industry expert interviews supplement this approach with practical experience and best practices.

The analysis of commonly used benchmarks identified in the interviews showed that well-established benchmarks and web server load testing tools were infeasible due to the lack of command line interface access in shared and managed hosting environments. Benchmarks relying solely on PHP and MySQL functions appear promising, but they do not account for all relevant aspects and cannot incorporate external factors. External load testing tools and page speed tools were found to cover aspects not considered by server-side benchmarks, such as caching mechanisms and networking. The research also highlights the need to evaluate website loading speed and back-end administrative interface performance separately, as they are both relevant to customers but are affected differently by caching mechanisms.

The integration of these data sources is recommended for a comprehensive and meaningful analysis. However, different user groups may attach different importance to certain aspects of hosting, depending on their intended use case. Therefore, a weighted evaluation of different benchmark data for different use cases is essential, leading to the development of a multi-layered approach to benchmarking WordPress-specific hosting environments. This thesis proposes to evaluate the performance of web hosting solutions by utilizing synthetic workloads across multiple metrics to simulate real-world scenarios for improved repeatability and consistency in performance measurements. The proposed approach leads to a thorough understanding of the performance characteristics of web hosting solutions, which can be applied, evaluated, and further developed in practical settings.