Get the latest tech news

Swift sucks at web serving or does it?

Contents - Benchmark method & apparatus - Benchmark results - Debugging the benchmark - Domain experts weigh in - Examining the load - …but… why is the success rate still weird? - Examining the benchmark tool - Conclusion A few weeks ago, Axel Roest published a simple web server comparison, that turned out to not be doing what it was thought to be doing. Figuring that out was a very interesting discussion that warrants a retrospective, to look at which parts were particularly helpful and which not so much.

⚠️ While it’s not necessarily wrong or unwise to evaluate how well a benchmark represents real world usage (whether generally or against specific cases), it’s an exercise that suffers from diminishing returns pretty quickly. Once I had figured out that kern.ipc.somaxconn directly controlled the problematic behaviour, the rest followed pretty naturally and quickly – I realised that what I saw in the system traces was in fact accurate, and that in turn revealed that the macOS kernel contains multiple surprisingly blatant and serious bugs (or at the very least dubious design choices, and lying documentation) regarding TCP sockets in non-blocking mode. It appears that the kernel bugs apply to Linux as well (although it’s not known if kqueue was in use there, as wrk also supports epoll and select), as the behaviour seems to be the same there, but I didn’t test that myself and it’s not completely clear from Axel’s posts.

Get the Android app