The experience gained from designing and implementing httperf clearly suggests that realizing a robust and high-performance tool for assessing web server performance is a non-trivial undertaking. Given the increasing importance of the web and the need for quantitative performance analysis, the importance of such tools will continue to increase as well. While httperf is certainly not a panacea, it has proven useful in a number of web-related measurement tasks and is believed to be flexible and performant enough that it could provide a solid foundation to realize macro-level benchmarks such as SPECweb. For those cases where httperf may not be the solution of choice, we hope that the experience and lessons reported in this paper will be helpful in avoiding the most common pitfalls in measuring web server performance.