The main purpose of this thesis was to research and select the most usable ways of improving web application scalability especially during high traffic peaks. The most concrete meter for the application scalability was defined to be the maximum throughput of the application.From the general web application performance guidelines, caching was seen as the most promising solution for improving the maximum throughput of the application. Because of this, a couple of advanced caching techniques were implemented into a real life application: caching of content fragments with Edge Side Includes (ESI) markup language and caching of user group specific HTTP responses with a Servlet filter.After implementing the Edge Side Includes support into the application, the maximum throughput of the application front page was measured to be roughly 3 times the original. Though this was considered to be a good improvement, it seemed clear that the rendering process of the front page response skeleton was unnecessarily heavy at this point, therefore limiting the true potential of the ESI solution.In order to improve the performance of the response skeleton rendering process, a Servlet filter was configured to cache the different response skeletons to a local cache for two seconds, thus reducing the need for processing in the application especially during high traffic peaks. With this improvement, the maximum throughput of the application front page now reached approximately tenfold value compared to the original version.In addition, the maximum total throughput of the application was measured under a realistic simulation of a high traffic peak before and after the optimization steps. As a result, the optimized application seemed to perform approximately 2.3 times better than the original.In the end, the thesis project can be considered successful due to the noticeable improvement in the application scalability. Furthermore, various additional improvement possibilities were discovered during the project, which could help to improve the application scalability even more in the future.
展开▼
机译：本文的主要目的是研究和选择最有用的方法来提高Web应用程序的可伸缩性，尤其是在高峰时期。对于应用程序可扩展性，最具体的衡量标准被定义为应用程序的最大吞吐量。从通用Web应用程序性能准则来看，缓存被视为提高应用程序最大吞吐量的最有希望的解决方案。因此，在现实生活中的应用程序中实现了两种高级缓存技术：使用Edge Side Includes（ESI）标记语言缓存内容片段，以及使用Servlet过滤器缓存用户组特定的HTTP响应。在实现Edge Side Includes之后为支持应用程序，应用程序首页的最大吞吐量据估计约为原始吞吐量的3倍。尽管这被认为是一个很好的改进，但显然在这一点上首页响应框架的渲染过程不必要地繁重，因此限制了ESI解决方案的真正潜力。为了提高响应框架的性能渲染过程中，配置了一个Servlet过滤器以将不同的响应框架缓存到本地缓存中两秒钟，从而减少了在应用程序中进行处理的需求，尤其是在高峰时期。经过改进，与原始版本相比，应用程序首页的最大吞吐量现在达到了大约十倍。此外，在优化步骤前后的高流量峰的真实模拟下，测量了应用程序的最大总吞吐量。 。结果，优化后的应用程序的性能似乎要比原始应用程序好2.3倍。最后，由于应用程序可伸缩性的显着提高，因此可以认为论文项目是成功的。此外，在项目进行过程中还发现了各种其他改进的可能性，这可能有助于将来进一步提高应用程序的可伸缩性。
展开▼
