When we started our company and got our first big customer, we had to choose between JMeter, SilkPerformer and LoadRunner. One was too limited and the others were too expensive and tied to Windows. So we decided to build our own tool and address all the shortcomings we saw, such as reporting, platform dependencies, and scale. We even had to change the name at one point because our first idea was just stupid. 
17 years later, we continue to invest heavily in our tool, XLT. We run 150 projects a year with it and rely on its extensive capabilities. And yes, it is open source under the Apache license and uses many other open-source components to which we also contribute.
This talk will take you on a journey from the initial ideas to the current state, including but not limited to expensive mistakes, life-saving feature decisions, interesting insights, and all the things we learned just by writing a big tool.
The tool thrives because it lives in the Java ecosystem. It saw migrations from Java 6 to 8 to 11 to 17, it uses libraries such as HttpClient, Apache Commons, and HtmlUnit. It benefits from all the investments the Java community makes, and it is fast, because Java is fast.
Of course, this is not a sales presentation, but rather a load and performance testing tool expedition. The talk will highlight real-world challenges and results. And maybe we will stop you from writing another tool, or perhaps the opposite.