HTTP/2 vs. HTTP/3 performance benchmark

แชร์
ฝัง
  • เผยแพร่เมื่อ 19 พ.ย. 2024

ความคิดเห็น • 30

  • @AntonPutra
    @AntonPutra  ปีที่แล้ว +1

    🔴 - To support my channel, I’d like to offer Mentorship/On-the-Job Support/Consulting - me@antonputra.com

  • @hashcode01
    @hashcode01 6 หลายเดือนก่อน +3

    Thanks for the benchmark
    Note that In scenarios where a website has numerous assets to load, like images, scripts, and stylesheets, HTTP/2 is better
    But for real-time applications such as streaming video or online gaming, HTTP/3's can provide lower latency and faster recovery from network disruptions

    • @AntonPutra
      @AntonPutra  6 หลายเดือนก่อน +1

      Okay, thanks. I'll try to include more scenarios next time.

  • @kamurashev
    @kamurashev ปีที่แล้ว +12

    Interesting, I was really surprised by the results. I would expect at least equal performance.

    • @AntonPutra
      @AntonPutra  ปีที่แล้ว +1

      Well it’s too early

    • @lucaskunze
      @lucaskunze ปีที่แล้ว +2

      I tested last weekend, and I got the same result: HTTP/2 outperformed HTTP/3 by at least 100ms. I also expected better or at least equal results, but unfortunatelly that was not the result 😐. Let's wait a little more.

    • @kamurashev
      @kamurashev ปีที่แล้ว +1

      @@lucaskunze thanks for the update!

  • @primux2007
    @primux2007 ปีที่แล้ว +10

    good job bro with this videos keep working i know you will success

  • @TheSergWolf
    @TheSergWolf 2 หลายเดือนก่อน +1

    КЛАД!!!!! Спасибо огромное

  • @bakhtiyargarashov805
    @bakhtiyargarashov805 ปีที่แล้ว +2

    Gives very nice idea about performance. Thank you

  • @SergeiKliuikov
    @SergeiKliuikov 6 หลายเดือนก่อน +1

    Kudos for doing such practical research to spot the differences, but this test needs some tweaking. The biggest issue is that it assumes a stable connection. If we test it with a network with high packet loss (like a 3G signal or shaky WiFi), we’d see a whole different story. Another factor is dealing with high-res images or videos, like in an online store; the results would be totally different. Another snag is relying on the full page load times. This metric is heavily dependent on JS speed and the rendering engine, which doesn't really reflect the capabilities of the protocols. Lastly, the tests should be run in a highly competitive environment, with lots of users loading pages simultaneously. We should also check the traffic speed from the proxy because for HTTP/3, it’ll show lower values.

    • @AntonPutra
      @AntonPutra  6 หลายเดือนก่อน

      Thanks for the valuable feedback. I'll take a look at how I can reproduce the scenarios that you mentioned!

  • @creo_one
    @creo_one ปีที่แล้ว +2

    Infrastructure not diversified enough, how do you ensure that you're not measuring your noisy neighbors?

    • @AntonPutra
      @AntonPutra  ปีที่แล้ว

      I ran this test at least 3 times, by recreating VMs..
      By the way you can find the ansible to reproduce in couple of steps.

  • @YogiIrfandi
    @YogiIrfandi หลายเดือนก่อน +1

    Thanks!

    • @AntonPutra
      @AntonPutra  หลายเดือนก่อน

      my pleasure. i'll refresh this one soon

  • @tinpa1756
    @tinpa1756 ปีที่แล้ว +1

    Hi, great video first of all. Secondly, I'm trying to recreate the test, but it seems like the playwright client doesn't use the HTTP/3 protocol. Whenever I try the endpoint in my Firefox browser I can see that it uses HTTP/3. Do you have any ideas why that might be?

    • @AntonPutra
      @AntonPutra  ปีที่แล้ว

      Thank you! No, unfortunately, it's been a while since I tested/recorded this video.

  • @Babbili
    @Babbili ปีที่แล้ว +2

    your videos are awesome

  • @KomentarSaya
    @KomentarSaya 9 หลายเดือนก่อน +1

    please do benchmark again in 2024,
    add openlitespeed in the list

    • @AntonPutra
      @AntonPutra  9 หลายเดือนก่อน

      ok :)

  • @home-uf3tg
    @home-uf3tg ปีที่แล้ว

    I have no experince with grafana. anyone cloud you please tell me how can I import test dashboard after I login localhost:3000 ?

    • @AntonPutra
      @AntonPutra  ปีที่แล้ว

      2 ways
      1. copy paste grafana dashboard id
      2. paste json directly

  • @bartoski-ag
    @bartoski-ag 11 หลายเดือนก่อน

    now do the same thing but with haproxy 2.9

    • @AntonPutra
      @AntonPutra  11 หลายเดือนก่อน

      sometime in the future

  • @ZelenoJabko
    @ZelenoJabko ปีที่แล้ว

    Myëtriks