The Backend Engineering Show with Hussein Nasser cover image

NGINX Internal Architecture - Workers | The Backend Engineering Show

The Backend Engineering Show with Hussein Nasser

00:00

The Cost of Parsing HTTP Headers in GNX

Each worker process is pinned to a CPU. And we never talk about why, why we have one worker per CPU because we really need that worker process to live in a in a core and never leave it as fast as possible. Why? Because there's something called contact switches. You switch the CPU off to do something else and you are put and hold as a process. That is expensive. It should be three that completely different protocol, it's quick. But the cost of this parsing, it lives right here.

Transcript
Play full episode

The AI-powered Podcast Player

Save insights by tapping your headphones, chat with episodes, discover the best highlights - and more!
App store bannerPlay store banner
Get the app