Stress Test: Page 2 - Enhance Control Panel

Stress Test

Adrien

AdamM I chose the ownership option. It made more sense in my financials, and I believe Litespeed is here to stay. Otherwise, the monthly license cost was too high.

Zoinkies

Adrien on the ownership version can you use it on as many servers as you want?

Adrien

Zoinkies In your dreams only ;-)
In LS reality, one license = 1 server.

Zoinkies I know they’re increasing prices soon.

This is bad news. I wasn't expecting that, considering how expensive it already is.

mendozal

Adrien @Zoinkies I recently moved to enhance self hosted (I was using a cPanel reseller account with Litespeed built in).

In your oppinion, Is it bad/less performant just to use OLS instead of LS? Considering the other option is Apache.

Adrien

@mendozal

mendozal In your oppinion, Is it bad/less performant just to use OLS instead of LS? Considering the other option is Apache.

I had some bad experiences with OLS in the past on RC, but they've made some improvements. It was with headers and UTM tag handling, for example, when it's appended to the URL for your marketing campaigns. The UTM tag wasn't passed.

OLS won't give you these features:

full .htaccess support and its caching
SSL Offloading
A more powerful cache engine
High performance mod_security engine
WordPress Brute Force Attacks Protection
Zero downtime server software upgrade
Version management
ESI support
Redis mass hosting

I have never conducted a serious benchmark between OLS and Enterprise, so I can't say if one is faster.

fbaye

mendozal NGINX competes with LS when configured properly.

mendozal

Adrien I see. I'm currently using OLS, but my client's websites are low traffic ones.

I had issues with .htaccess support and Wordfence wordpress plugin. But for now is manageable.

I just hope I won't hit any of the other missing features. I don't want to pay LS's per server license since it's completely in opposition to Ehnance's per website policy.

fbaye

I managed to get 18k clients per second with Enhance on PHP7.4.

https://imgur.com/a/kHyAOsJ

In fairness, on my cPanel stack I did get 27k clients per second before.

AdamM

Now I would like to show you something why I think that something is wrong with Enhance or that LiteSpeed itself is not working as it should.

I created a test environment in 10 minutes to show you the results:

ubuntu 22.04 in Vultr - HF
PHP - installed to simply "sudo apt install php8.1-fpm"
Caddy Web Server - https://caddyserver.com/
simple php file with hello world
kernel 6.2, if it makes a difference?
go language 1.20

Here a simple Caddy configuration for PHP support in a few lines (without cache or other settings)

PHP file

A simple http test from a worldwide location - I have a test server in Poland

CPU consumption during the test - not visible

And now for the best, a benchmark using loader.io

Do you see this result?
Success Response - 131158 requests / per minute 😳
Average response 114ms

This result is only for a server with 1 cpu and 1gb ram
Now can someone explain to me why this is possible here? Is it simply that Caddy handles PHP so well?
It seems so... in future I would like to use Caddy instead of LiteSpeed because it is efficient, free and opensource

What do you think? Why is this result possible in Caddy and not in LiteSpeed without cache? Unless Caddy has some cache mechanism that I do not know about? But I should also mention that I changed the code of the PHP file for the test and after refreshing the browser the new content was already served, without any delay...
In litespeed with no cache enabled the response is up to 5s and uses all the CPU power, this is not normal....

What are your thoughts?

fbaye

AdamM Send me your loader.io settings/ config

Adrien

AdamM I'm concerned.
Honestly, I thought containers were helping with WP performance by isolating one another. It's also my very first experience using LiteSpeed Enterprise. I was really hoping for better perf then with RC OLS. After noticing my results were off, I hired the LiteSpeed team to check my AMD 7950X. I was hoping to join friends for a contest but declined because I felt something wasn't right, but I don't know what.
I can't even handle 3000 users before my test fails.

fbaye Here are my loader.io settings:

Hopefully we can find out and improve the perf.

AdamM

fbaye I only add Headers - Accept-Encoding - gzip,deflate

fbaye

AdamM I mean take a screenshot of this screen https://imgur.com/a/zWBb525 I want to make sure our tests are the same.

AdamM

Zoinkies

AdamM Check your tidio chat 🙂

fbaye

https://imgur.com/a/D9n5fVh This is my configuration matching your tests with LSE. Ignore the initial high TTFB had rate limiting enables 🙂

AdamM

fbaye I understand but we are talking about something else, you use cache right?

AdamM

fbaye What our machine for such a good result without cache.? How much cpu and ram?

fbaye

AdamM No cache

AdamM

fbaye so is a very good result

AdamM

Adrien of course you use cache litespeed ? A lot also depends on the website size.
Can you provide the final result?

Adrien

AdamM I did. Maybe you missed my post above.

Adrien RC 5950X. 0 - 5000 clients over 1 min
Results: https://bit.ly/2XPCkGk

Enhance 7950X. 0 - 5000 clients over 1 min
Results: https://bit.ly/3IXfUYJ

At the moment, the test failed on Enhance. Yes, I'm using Litespeed cache plugin.

« Previous Page Next Page »