WordPress 5.0 vs. 4.9 – The Performance Test

Jan Hornung
8 Min.
WordPress  5.0 and Gutenberg_performance comparison
Last updated on

The update to 5WordPress .0 is just around the corner. However, the exact release date depends on when the new WordPress -editor alias "Gutenberg" is ready for the WP core. Many WP users are worried that there will be problems with Plugins, Themesand the performance of their site WP after the update. For this reason, we took a closer look at the update and measured whether 5WordPress .0 really does lead to the dreaded load time losses.

Update: The article refers to a test that we conducted in the alpha phase of the Gutenberg-Plugins. At December 6, 2018 5WordPress .0 has been released with the Gutenberg-editor. In our E-Book (only available in German) you will learn what the new editor means for your business and how you can make your website Gutenberg prepare. Whether Gutenberg can keep up with established page builders by now, you will learn in this article of Tariq Baig.

Gutenberg and 5WordPress .0 e-book

The coming WordPress -Update to version 5.0 has a main actor: the so-called Gutenberg-editor. However, after integration into the WP core, this project name will be history to allow new WordPress -not to confuse users unnecessarily. At least that's the plan. But let's first come to the most important question: What exactly is the Gutenberg-editor anyway?

Gutenberg should WordPress revolutionize

If you've already spent a lot of time with the Gutenberg-basics familiar, then jump directly to our WordPress 4.9 vs. WordPress 5.0 performance comparison.

The major update to 5WordPress .0 will Gutenbergsoon replace the classic TinyMCE editor, which WP users have been using for over 10 years to sites create and edit articles without great HTML knowledge. However, if you want to create more complex layouts, the TinyMCE editor without HTML knowledge and shortcodes quickly reaches its limits.

This hurdle should Gutenbergbe removed to make it WordPress even more accessible and user-friendly. The basic principle behind the new editor is a series of blocks that provide special functions. For example, there are blocks for headings, text, columns, quotations, lists, images, videos, widgets, and and and.

WordPress 5.0 brings more JavaScript and uses the WordPress REST API

The new editor brings another change: JavaScript: "Most WordPress developers are pretty comfortable with PHP and front-end development, though Gutenbergblocks are heavily JavaScript based," writes WP developer Rich Tabor in his Beginner's Guide to Building GutenbergBlocks. It's not for no reason that WordPress co-founder Matt Mullenweg has already introduced the WP community to State of the Word 2015 the homework "Learn JavaScript deeply!

Due to these fundamental changes through 5WordPress .0, we are curious to see to what extent Gutenberg's use of JavaScript and the WordPress REST API on the performance of various WordPress sites is having an effect. This is because a lot of computing effort is shifted from the web server to the client, which should lead to a shorter loading time under load. But we will come to that in a moment.

WordPress 5.0 is a step towards pagebuilder

Within the WP community there is often talk about WordPress offers the users "What you see is what you get" (WYSIWYG) However, the reality often looks different. Those who work with WordPress knows that on his you donsite 't always get what you saw in the back end.

Gutenberg is supposed to close this gap and WordPress bring a big step closer to WYSIWYG. To this topic and the new possibilities by the Gutenberg-editor I can give you the Lecture by Morten Rand-Hendriksen and Mullenweg State of the Word 2017 from WordCamp US recommend.

The fact is, it Gutenbergwill initially only be introduced at the editor level, but this is only the beginning. Because Mullenweg's vision is to make it completely WordPress block-based in order to be able to keep up with the competition from website construction kits like Wix, Medium and Co.

The next step - so what comes after what you just saw, when we finished the editing experience, is the Gutenberg-based site customization. - Matt Mullenweg, State of the Word 2017

For the future, this could lead to a paradigm shift inWordPress , away from the familiar structure Pluginsand Themestowards a single huge pagebuilder.

WordPress 5.0 raises concerns due to "Gutenbergcompulsion

The fact that Gutenberg with WordPress 5.0 becomes the default editor, that is, the WordPress -is virtually forced upon users, not everyone likes it. Plugin- and Thememanufacturers must, for example, make their products Gutenberg-compatibility, adapt them if necessary or rethink them completely.

Also for operators of WordPress sites questions are increasingly coming up: How does this affect my business? Power WordPress 5.0 my site slower? Will mine Pluginsstill work? Will the update bring me something if I don't use the editor at all?

For this reason, we will now move away from the discussion about Gutenbergblocks and the like and turn our attention to the following questions:

1. is there a performance difference between 5WordPress .0 and 4.9?

2. in view of the current uncertainties, should you update to 5WordPress .0 at all?

Load time comparison: WordPress 4.9 versus WordPress 5.0

For our systematic comparison of the two WordPress versions, we have chosen two key performance indicators: the page load time on a single call and the behavior of the pages sites under load.

We have prepared four sample pages for the load time comparison, whose load times we have systematically compared under WP 4.9.4 and WP 5.0 (in the 5.0-alpha-42852). For this purpose we have developed the tool Webpagetest.org is used.

And to anticipate, there are no serious differences.

Test 1: 5WordPress .0 tends WP site to slow down an empty

In the first test we have checked the loading times of a completely empty under WP site WordPress 5.0 and 4.9 were compared. The result: The Time To First Byte (TTFB) and the perceived load time of the (Load site Time) have been compared with WordPress 5.0 slightly increased. The total loading time has slightly decreased. At the latest when looking at the cached results, this difference is completely equalized.

The conclusion of this first test: The differences in the measured values are probably variations in the measurement itself and not systematic differences in loading time.

Test 1: Empty loading timeWP site
On the two tested DEMO BOXES the ThemeTwenty Seventeen is active and the ThemesTwenty Sixteenand Twenty Fifteenare installed. There are none Pluginson the site . For both sites the RAIDBOXES Cache active.

Test 2: How does an online shop behave under 5WordPress .0?

In the second step, we used the same test setup for a sample shop. Here the start page of the shop was called up in each case. In additionWooCommerce, more than 50 more were Pluginsinstalled on the test page, more than 45 of them were active during the test.

The shop, which is equipped with products and functioning payment gateways, also showed no major differences in important performance indicators during the test.

The TTFB, Load Time and Webpagetest Speed Index vary only slightly. Here, too, the differences become even smaller with cached calls.

est 2: Example shop under 5WordPress .0
Especially in the cached state you can clearly see that the differences between the two WordPress versions are negligible.

Conclusion for shop owners: With a realistic and quite extensive shop we could not find any systematic differences between WP 4.9.4 and WP 5.0 - a good sign!

Test 3: How does a blog behave under 5WordPress .0?

Even in our test with a typical blog - with extensive social media connectivity and more than 20Plugins, like Yoast SEO, WP Smush, Akismet and Contact Form 7, there are no really big differences in the important load time characteristics: Server response time, perceived load time and total load time differ only slightly between the WordPress versions.

Conclusion: In view of the Speed Index and the Start Render Event, the blog seems to load a little bit faster under WP 4.9.4, but all in all there are probably no significant differences.

Test 3: Example blog 5WordPress .0
When testing the loading time of the blog, the differences seem to be a bit bigger, but really impressive differences do not show up.

Test 4: What about a corporate site below 5WordPress .0?

In the company page we tested, interesting differences are revealed for the first time: Although TTFB, the Start Render Event and the Speed Index are quite close to each other, there is a big difference of 1.8 seconds in the perceived total loading time.

However, this value can be explained by an incorrect resource. If this is removed, the load time values are equalized again. So the conclusion here is the same: The company site we tested with Custom Themeand 9 (including PluginsWPML, Advanced Custom Fieldsand a tool for landing pages) shows no relevant differences between WordPress 4.9.4 and WordPress 5.0.

Test 4: Example company page 5WordPress .0
The difference in load time equalized after we removed the faulty resource. This makes the above test a perfect example of how susceptible load time measurements are to even the smallest error, such as a faulty image.

Interim conclusion: 5WordPress .0 is not a performance eater

Our first analyses with an alpha version of WordPress 5.0 show that the new WP version will probably not be a performance eater - at least not in the frontend. No matter if with or Gutenberg: Relevant differences do not show up after our measurements. From a performance point of view there is nothing to be said against an update WordPress 5.0.

On the contrary, if you look at the results of our load test, the update could well be worth it.

Important: Of course, these results may still change over time. Also Pluginsand Themesand their compatibility with the new WP version will play a decisive role.

Test 4: 5WordPress .0 shows better load handling

Besides the loading time of onesite , it is also important how many concurrent accesses it can site handle before it runs into a timeout. And this is where we can see some really serious differences between WP 4.9.4 and the version of 5WordPress .0 we tested.

For our test we have created an empty one with WP site the tool Loader.io is loaded with 500, 750 and 1,000 uncached calls per minute and the result is remarkable. While the site one with WP 4.9.4 already goes down to its knees at 750 requests per minute and shows an average response time (i.e. Time To First Byte) of 4.9 seconds, the site one with 5WordPress .0 is much more resistant to stress.

Even with 1,000 requests, the average TTFB of the site still - not good but stable - 863 milliseconds. This is less than a ninth of the value of the site 4WordPress .9.4.

Test 5: Load test empty WP site

To be honest, we could hardly believe this difference, but even during the third, fourth and fifth runs the differences remained stable.

This result surprised us so much that we did the same test with the Apache Benchmark Tool in the backend. Again, the result is astonishing: The tests site under WordPress 5.0 can process 50 percent more requests than the site WordPress 4.9.4.

Due to the optimization of the source code in 5WordPress .0, some resources are saved on the server side. This means that the available resources can handle more requests at once.

Conclusion: 5WordPress .0 has surprised us positively

The final extent to which WordPress 5.0 changes the performance of yourssite , will only show in the beta and full version. But one thing is certain: WordPress 5.0 runs much more stable under load than version 4.9.4 in our test.

From the perspective of load compatibility, we can therefore only recommend the update to 5WordPress .0 at this time. However, as already mentioned, many things can still change before the release, including these test results. 

In addition, we have focused completely on performance in these tests. As soon as the 5WordPress .0 beta comes out, we will also look at compatibility with Pluginsand Themesand general functionality of the Gutenberg-editor. Only then can we give a clear update recommendation. So we keep an eye on the topic and hopefully we can provide you with fresh data soon.

Have you tested 5WordPress .0 or that Gutenberg-Plugin already tested? Then please leave me a comment, I'm looking forward to further experiences!

Related articles

Comments on this article

Write a comment

Your e-mail address will not be published. Required fields are marked with * .