TechEmpower just released their latest round (19) of their server middleware benchmark. You can check the results here: https://www.techempower.com/benchmarks

Very interesting to see the results. I am really wondering how such large differences can happen. Are JVM based frameworks misrepresented because of missing warmup phase? I doubt it. Still crazy to see more than 5x differences…

@Draget Actually, there is a warm up phase before running the benchmarks

Some Frameworks probably get a penality for being more dynamic about routing options, like parsing configs and states for each request while others compile to static handling code. And yet I think some of those performance differences should be incorperated into test-suits of said frameworks and push them towards better performance.

I’m missing the disclaimer that says “your database is usually the bottleneck, so please fix that shit before looking at this”

At first thought, it can be strange to compare frameworks only on performance as we use them for more added values than that. Still, this cannot be ignored. This will maybe not be suited to find the best framework for a specific project but it can help to avoid the one that can be a bottleneck. I think that the best way to approach the results is to read the

motivation

section and to use filters to remove data noise. thanks @jaguililla