NGINX Internal Architecture - Workers

In this video I explain the NGINX internal process architecture, how NGINX spins up multiple processes per core, how each process is pinned to a CPU core for minimum context switching, how NGINX accepts connections , parses requests and talks to the backend.
Get my introduction to NGINX udemy course
nginx.husseinnasser.com
Fundamentals of Networking for Effective Backends udemy course (link redirects to udemy with coupon)
network.husseinnasser.com
Fundamentals of Database Engineering udemy course (link redirects to udemy with coupon)
database.husseinnasser.com
Introduction to NGINX (link redirects to udemy with coupon)
nginx.husseinnasser.com
Python on the Backend (link redirects to udemy with coupon)
python.husseinnasser.com
Become a Member on KZbin
/ @hnasr
Buy me a coffee if you liked this
www.buymeacoffee.com/hnasr
Arabic Software Engineering Channel
/ @husseinnasser
🔥 Members Only Content
• Members-only videos
🏭 Backend Engineering Videos in Order
backend.husseinnasser.com
💾 Database Engineering Videos
• Database Engineering
🎙️Listen to the Backend Engineering Podcast
husseinnasser.com/podcast
Gears and tools used on the Channel (affiliates)
🖼️ Slides and Thumbnail Design
Canva
partner.canva.com/c/2766475/6...
Stay Awesome,
Hussein

[@hnasr]

Get my introduction to NGINX udemy course nginx.husseinnasser.com

[@RajnishAhuja]

Hussein, Do you have a course on istio and certificate management ( By you or somebody you respect ) ?

[@tempaccount8256]

Hussein, your contents are always awesome, with a lot of information. I was wondering if you can tell us about your past experiences and how you get to the place you're now, and also what's your current stack? And more😁😁, We will be lucky😅

[@shubhamdhal689]

First Comment on a Hussein Nasser vid :) One thing to note is how fast KZbin is able to serve the video in my feed. I think I started watching this almost as soon as Hussein uploaded the video(it was saying zero views, but ofcourse that counter is eventually consistent and is cached at multiple layers). KZbin recommendations is truly a feat of engineering

[@franpereyra5459]

It would be great to see an explanation like this, about Apache web server, in order to understand the main differences with Nginx (pros and cons). Amazing explanation, thank you!

[@hnasr]

Great idea

[@nickkman90]

Fabulous lecture! Thanks Hussein.

[@nikhilsinghal1315]

I work on nginx modules a lot. You should also discuss the moduler way of adding new code in nginx. The code architecture of nginx is pretty neat.

[@rio512hsu]

Any good reading material recommended?

[@krishnamohanyerrabilli4040]

👏 thanks for this, clearly understood what's happening behind the scenes

[@shreemaan-abhishek]

Oh hey!

[@blevenzon]

Thank you for the clear and concise explanation 😊

[@habibicoding]

Very interesting, good job, shater/شاطر!

[@KuldeepYadav-jn3ld]

Could you please have a small session about Public Suffix List ?

[@TylerRake141]

I would love to watch more stuff like these, it could be a series like e.g. IIS, Apache architecture etc

[@rehaanmazid2435]

Hey @Hussien, When a NGINX server is set up in docker. What happens to the worker process and core mapping? I assume it will not map to the host machines cpu cores. What happens in this case?

[@SB-qm5wg]

nginx is one of those things I've barely admin'ed at jobs but never put time in to master.

[@AjithkumarSekar]

This is great. Can you also talk about CGI (Common Gateway Interface) and why we need it?

[@faizsyed658]

Do you want internal architecture of CGI? If you just wanna better understand about cgi, there are many good videos from other channels already

[@dimtass]

A good strategy, especially to bare metal embedded servers is to use SMP IRQ affinity and map the netif IRQ to a single core. Then only use the rest cores to handle requests. This will improve the performance a lot.

[@thescubabuba6470]

Is there a way to calculate how many connections a worker can handle simultaneously?

[@hnasr]

I think this is a very good question and needs low level kernel level and logging to see how much latency is added with 1 , 2 and 3 and then 4 threads. And then seeing how many requests can be processed with each configuration

[@achrefnabil2463]

I think if you wanna calculate the maximum number of connections per worker process you should divide the system file descriptor limit by the number of worker processes. For example, if the file descriptor limit is 1024 and there are 4 worker processes each worker process can handle up to 256 connections simultaneously

[@manojBadam]

How does file descriptors come into picture for handling more connections?

[@TylerJusCodes]

Hey .. long time my friend. been awhile I visited here.

[@seanknowles9985]

How to build a http3 client, I am struggling to get my head around why some http libraries are still http1 and why they've not been updated? (this could be interesting content)!

[@raghavgupta1799]

This topic should be in the Indian Engineering courses as a separate subject.

[@mohammedabulsoud1819]

so goood

[@abelsimon5308]

planning a course on envoy as well? (have seen your TL-DR about the proxy)

[@hnasr]

made a video on envoy but im not satisfied with it. need to redo it.

[@abelsimon5308]

​@@hnasr maybe a udemy course if you fancy would be even better I've thoroughly enjoyed the nginx one. It does also help with the bigger picture when having multiple layers of proxies. E.g a project I'm working on right now, we're having an HA proxy and two envoys for egress. (one sidecar with custom listeners set up for port forwarding, and one egress in a DMZ to fulfil these forwards)

[@erlangparasu6339]

is nginx open source?

[@KhanKhan-ym4mo]

u shud take some presentation / communication skills training, ur videos are very monotonous and uninteresting, despite having valuable content. i hope you will take it positively and improve urself.