Question Why is CPanel so much faster?

[B4c4rd1]

Hi guys,

I do not want to have a principal discussion here. But I was wondering what exactly the difference is to the Plesk Panel. I have created 2 identical virtual machines. Same database, same operating system and same hardware settings. In each case CPanel and the Plesk Panel were not configured in detail.

In CPanel I installed the package PHP FPM with OPCache. Used software for testing: Shopware 5

Now to the test!

The TTFB makes a difference of 180ms!

• With CPanel the call of the login page of Shopware 5 needs only 20ms.
• With Plesk Obsidian and ionCube it takes about 200ms.
• Without ionCube about 150ms.

The ionCube encoder was the only clue that slowed down the system at Plesk noticeably.

But I am interested what makes CPanel different to Plesk ???

Has anyone already dealt with this topic? It would be really good if you could achieve the same speeds with Plesk. Maybe one of you knows what the difference is. Is EasyApache the reason?

I look forward to your answers!

 

[Bitpalast]

Your numbers are all the same, where is the difference?

 

[B4c4rd1]

Ah, please excuse this mess. I have just corrected the first post...

CPanel FPM with OPCache PHP 7.4:

TTFB = 20~25ms

Plesk FPM with OPCache PHP 7.4:

With ionCube TTFB = 180~200ms
Without ionCube TTFB = 150~ms

 

[Bitpalast]

If ionCube loader is needed to run the software why is it not needed on the other system?
Is the same PHP version used on both?
Have you checked with a tool like the Pingdom speedtest where the websites needs most of the time to respond?

 

[B4c4rd1]

@Peter Debik the test was the Chrome Browser. These were virtual machines installed locally. The ionCube Encoder is not automatically installed with CPanel. Which was not the case with Plesk. I found out later that the ionCube encoder is about 50ms slower. They are not accessible from outside. But it's more about why Plesk is slower than CPanel.

Shopware 5 is a very good test environment to find such differences.

 

[Bitpalast]

If the website setup is not really the same it is difficult to tell why one is responding slower than the other. Webserver, PHP all being the same, there should not be any difference. The panel software has absolutely no influence on the delivery speed of a page. It's the configuration of the web server maybe, but that would not match that you are saying this is the same in both installations. It probably is not. Maybe you can compare your web server configuration files? Also, if one software needs ionCube loader, it must be slower than the one that does not use ionCube loader, because the decryption process and additional hard disk access needs more time.

 

[B4c4rd1]

I have already written at the beginning that I do not want to have a discussion of principles.

Let me summarize once again:

- Both systems have the same hardware.
- Both systems have the same operating system in the same basic configuration.
- Both systems use PHP 7.4 as FPM with OPCache.

And now to the same question, what makes CPanel different from Plesk that the difference is so significant?

I want to know what the blockers are. For example the ionCube is one of them.

The panel is mainly responsible for which packages are installed and which settings are set.

So where does the difference of 130-150ms come from ?

I would be very interested to know what the trigger is at Plesk. I would like to achieve the same values with Plesk. Maybe you have already gained some experience with it?

 

[Bitpalast]

I don't need a lecture about what the ionCube does. I did not ask for that either!

The panel is mainly responsible for which packages are installed and which settings are set. Please don't tell me anything else, because that would be nonsense.

Glad to see that you already know it all. Good luck with your further analysis.

 

[B4c4rd1]

is someone mad at you? Very professional @Peter Debik !

I wasn't finished with my post (By mistake I have sent the post with my cell phone). I did not want to send it this way. Nice to see you're about to take this in.

At this point this should not go out to Peter. So please delete this Peter!

 

[enerspace]

Leave Peter on the left. We will try to solve the problem together with you.

Plesk has never been interested in being compared with CPanel. Most of the time the Plesk community reacts negatively to such requests, although it would offer added value to learn from both systems.
We know that Plesk goes a different way, which we do not always like.

We have also been doing research for some time on why Plesk is slower. We have also made positive progress, but we do not want to share this with everyone. Therefore, we find the topic you touch on here very interesting.

If you are interested, please write us a private message.

we speed up Plesk with you

 

[ChristophRo]

Most likely the low TTFB on CPanel simply comes from disabled output buffering

On Plesk this is enabled (at least if you let Apache2 handle the php-fpm stuff and use nginx in proxy mode) as the proxy_fcgi module will buffer everything from the php-fpm process till the script has completed. (php-fpm itself will most likely stream it's output in 4kB chunks)
This is due to the fact that you will need at least Apache2 2.4.32 and it's included proxy_fcgi module, in order to disable output buffering on that level.
So there is no way for Plesk to enable this on the majority of it's supported operating systems, except if they would start to ship their own Apache2 packages.

And yeah, this can have a huge impact on perceived page speed and loading times.