speed optimization

Will HTTP/2 Make Your WordPress Site Faster?

What’s HTTP/2?

With out boring the pants off you, HTTP/2 is an up to date and more efficient method of delivering website online elements from server to browser. There are 3 circumstances:

  • Browsers need to help it – most of them do now.
  • Servers need to help it. Many do, ask your host about it. In the event that they don’t, using Cloudflare will enable HTTP/2
  • Your website has to use HTTPS

Now that it’s turning into increasingly widespread, most articles on the topic make sweeping guarantees of quicker performance, “just like that”, simply by enabling it. But there are fewer articles which truly back up these claims with check outcomes.

I lately transformed a few sites from HTTP to HTTPS and determined to take the opportunity to see what difference, if any, enabling HTTP/2 made.

BONUS PDF: Demystify the jargon! Download this guide to widespread velocity optimization terms.

Check setup

The sites I tested are hosted within the US with Siteground’s Develop Huge plan, utilizing SSL certificates from Let’s Encrypt, that are offered by Siteground without spending a dime.

Round 1

The first website I examined has a blog-style homepage with several pictures and some external content within the type of two YouTube videos and one Vimeo video. These are the specs:

page size and requests for homepage

Benchmark with none optimizations

The first set of exams have been finished from a US-server, with none caching or optimizations.

Winner:  Too close to call.

The outcomes have been incredibly comparable excluding the one outlying outcome on the HTTP/1.1 aspect which was brought on by the third social gathering information loading slowly. So simply by enabling HTTP/2, there’s not any astounding enchancment.

This isn’t actually shocking on this case for a few causes:

  1. This specific web page has a variety of external content on it, and that may tank your performance every time, HTTP/2 or not. It’s going to all the time trigger more fluctuation in loading occasions. There’s no repair in addition to removing it.
  2. There are solely 18 belongings loaded from the origin server (i.e. from the location’s personal domain quite than 3rd events),  so it doesn’t actually benefit from HTTP/2’s multiplexing, that is, the power to serve many information over one connection, enhancing efficiency.

Caching and mixing belongings

Next I activated page caching together with combining CSS, JS and Google fonts (using WP Rocket). Combining these asset varieties leads to fewer HTTP requests, that is, fewer particular person information to be downloaded.

Homepage - caching and combine

Winner: HTTP/1

To some extent that’s expected because of the widespread knowledge of mixing information being a greatest apply for HTTP/1 but not essentially for HTTP/2. Nevertheless this will definitely range from website to website, and as you’ll be able to see, the margin of difference is just not large,  so it is best to undoubtedly check it for yourself.

Caching and minifying belongings

In this set of exams I used caching with only minification of belongings, however with out combining:

homepage -caching and minification

Winner: HTTP/2

This is anticipated since combining belongings isn’t useful for HTTP/2 performance, and detrimental to HTTP/1 efficiency. The margin of victory here is more vital than within the earlier set of exams using combining.

Round 2

For the subsequent set of checks I decided to attempt a special type of web page on a special website – one without any exterior content, and with a better variety of requests:

Site 2 - page specs

Benchmark with none optimizations

Site 2 - no caching

Winner: HTTP/2… just

HTTP/2 has a slight edge here, doubtless because there’s more inner content which may reap the benefits of multiplexing. By comparing the waterfall charts under, you’ll be able to see, file by file, how much quicker they are on HTTP/2.

The inexperienced section of the bar represents connecting time and the beige section is the blocking time. Per GT Metrix, blocking is the time the request spent ready in the browser’s queue to start out the request.

Both blocking and connecting are enormously lowered on every request with HTTP/2.

http/2 vs http/1.1 - request timings

Caching and Minifying

site 2 - cache and minify

Winner: Lifeless heat

This was a bit shocking. I might anticipate HTTP/2 to win right here, because the lack of combining belongings ought to favor HTTP/2.

Caching and mixing belongings

Site 2 - cache and combine

Winner:  HTTP/1.1

This matches the Round 1 outcome on this class – combining being an efficient method for optimization on HTTP/1.1

General, a combined bag of results right here, with out displaying a clear benefit for HTTP/2.

Round 3

One of the assumed benefits of HTTP/2 is that it’s extra efficient in larger latency situations. Meaning situations the place it might sometimes take longer for knowledge to be downloaded. Bodily distance between the server the place the location lives and the consumer requesting the online web page, increases latency. So, to test this I did a set of exams from the China server location on GT Metrix, which would definitely improve the load time when a CDN shouldn’t be being used.

China: No caching

No caching - China

Winner: HTTP/2

Right here we see the most important distinction between the 2, with HTTP/2 being virtually a second quicker, confirming the speculation that it performs higher in high-latency conditions.

China: Caching and minifying

Cache and minify - China

Winner: HTTP/2

HTTP/2 has a clear benefit on this state of affairs.

China: Caching and mixing belongings

Cache and combine - China

Winner: HTTP/1.1

Perhaps surprisingly, once the web page is optimized for HTTP/1.1 it becomes just as effective as HTTP/2, even in this state of affairs.


Time To First Byte

Some individuals, and velocity testing instruments have been using Time To First Byte (TTFB) for a long time as an indicator of how briskly a website is. In the HTTP/2 world, this isn’t necessarily going to be related. You’ll persistently see an extended TTFB on HTTP/2 enabled websites, and there are expected reasons for that.

Many people, as I was in these exams, can be shifting from sites using HTTP/1.1 without SSL, to including an SSL. Many hosts are routinely enabling HTTP/2 for SSL sites.  Because the HTTP/1.1 version didn’t have use HTTPS, the velocity comparison isn’t  utterly apples-to-apples. The SSL negotiation does add one thing to the TTFB. It will have been a more accurate check to match HTTP/1.1 with SSL and HTTP/2 with SSL, to be able to set up how much of that longer time is simply as a result of HTTP/2.

For a more technical rationalization of why TTFB will not be the most effective metric to take a look at with HTTP/2, I recommend this article on the Akamai weblog.

And on the Akamai boards, a consumer provides an incredible analogy, evaluating the distinction to the previous vs new ordering system at McDonalds. Primarily, the same processes need to occur and you need to do some waiting in both case, but you’re simply ready in a unique a part of the process. The takeaway is that, although the TTFB might appear longer with HTTP/2 it’s not related because different elements of the process are a lot quicker now.
Let’s examine these two exams:

Within the screenshots under we see that the TTFB for the HTTP/1.1 outcome is definitely rather a lot quicker, however HTTP/2 still ends up being faster general,  and the primary paint is quicker, comparatively speaking.

Here’s the HTTP/2 outcome:

HTTP2 - relative TTFB and first paint

The primary request takes 2.14s

The first paint relative to the start is 1.35s

Here’s the HTTP/1 outcome:

HTTP/1 - relative TTFB and first paint
The primary request takes 1.68s

The primary paint relative to the start is three.34s

Perceived Velocity Metrics

With the relative velocity displayed above, you’d in all probability anticipate metrics like First Render and First Contentful Paint to be a lot quicker with HTTP/2. I did not find this to be the case. Evaluating these metrics throughout quite a few checks produced a real combined bag of results.

Evaluating the Contentful Paint metric of the above two checks could also be shocking.


HTTP/1 - contentful paint and DOM loaded

Contentful paint: 3.3s
DOM loaded: four.8s


http_2- contentful paint and DOM loaded

Contentful paint: 3.5s
DOM loaded: three.7s

The overall load time is 1 second quicker for HTTP/2 but the Contentful Paint doesn’t show that a lot of a distinction. There’s a a lot smaller hole between Contentful Paint and DOM loaded which is probably going why the general Velocity Index is best. After viewing the cellular results under, this could possibly be deceptive.

What I noticed in my checks was that First Contentful Paint wasn’t significantly quicker, if in any respect, with HTTP/2.

Confusion Abounds

What you might be noticing at this level is that  measuring HTTP/2 efficiency up to now is fairly confusing and your mileage will range based on what you are feeling the essential metrics are. Commonest users gained’t really be wanting at the nuances right here and the most typical testing instruments – GT Metrix, Pingdom, PageSpeed usually fall brief in relation to HTTP/2. This is going to make it troublesome to completely perceive what’s occurring on your website, until you’re already really educated on the topic.

There were some situations the place HTTP/2 produced a quicker First Contentful Paint even with an extended load time. For example, in comparing these two studies, we will see that the general load time is 0.2s longer on HTTP/2, but the First Contentful Paint is zero.4s quicker on HTTP/2.

Word that the TTFB can also be virtually twice as lengthy with HTTP/2 however this doesn’t negatively impression the result because the efficiency of downloading the belongings more than makes up for it.

However what was extra typically the case was that there wasn’t an overarching development as to which carried out higher for perceived velocity. Now there could possibly be different causes for this earlier than leaping to the conclusion, “HTTP/2 is bad”.

For example, an enormous issue is the optimization of the server itself, specifically for HTTP/2. For those who’re really into the technical features of this, I like to recommend you take a look at this Twitter change on the subject, since it really demonstrates how the server prioritization settings play a task in all this.

The typical WordPress consumer nevertheless might be saddled with whatever their internet hosting plan supplies. On the time of publishing, Patrick Meenan (creator of webpagetest) had just released this WordPress theme particularly to check how nicely your server is optimized for HTTP/2. I haven’t yet had an opportunity to attempt it.

BONUS PDF: Demystify the jargon! Obtain this information to widespread velocity optimization phrases.

What about cellular?

Issues are even muddier if you begin testing cellular. This would wish much more testing on numerous sites, but there doesn’t appear to be any actual benefit to HTTP/2 on cellular yet.

In these 2 checks I compared the uncached page on HTTP/1.1 and HTTP/2 utilizing the cellular testing at webpagetest.org, and simulating a quick 3G connection. That is slower than most US/European customers would experience since we principally have fast 4G. But sometimes the slower the connection, the bigger the differences in optimization you possibly can see. Optimization, or lack thereof, impacts users within the worse circumstances, not those with the most effective.

The general load time is significantly quicker with HTTP/2, but HTTP/1 has a clear edge in displaying content earlier, as seen in the Start Render occasions:

Visually comparing the loading means of HTTP/1 and HTTP/2 exhibits that the previous begins to render more significant content material, in this case, the title of the publish, a lot sooner. But comparatively talking it takes longer to get to visually full from that point.

Whereas on HTTP/2, the consumer is stored waiting longer for anything meaningful (not sometimes good for perceived velocity) but then the whole lot masses in a lot quicker from that time. This echoes the results above from GT Metrix desktop exams the place the hole between Contentful Paint and DOM loaded or onload was a lot smaller on HTTP/2 than HTTP/1.1.

HTTP/1: The title exhibits at 9 seconds, the pictures at 13 seconds:

HTTP/1 - mobile - filmstrip

HTTP/2: The title exhibits at 14.5 seconds, the pictures at 15.5s:

HTTP2 - mobile - filmstrip

Given that every article you’ll learn will inform you how good HTTP/2 can be for cellular, that is shocking. I feel the idea is that cellular customers face greater latency issues, and in desktop checks, HTTP/2 did better in such circumstances. However cellular units produce other points, reminiscent of packet loss, and in those instances, HTTP/2 does worse. I don’t know the right way to check for packet loss, so I can’t say if that’s the rationale for these results.

An essential caveat is that in these exams I haven’t tried to optimize the location specifically for HTTP/2, using out there methods like server push. It’s potential that when you really optimize specifically on this method, the results might be much totally different. This may be the subject of another submit 😉


The commonly held knowledge of  “just turn it on and your site will be faster” isn’t necessarily true. It’s essential to keep in mind that these outcomes are based mostly on testing solely a couple of sites and cannot be thought-about consultant of the whole internet 🙂 However, as with something related to efficiency, HTTP/2 is not any silver bullet.

It’s definitely going to be the inspiration of the online going ahead so I’m undoubtedly not suggesting resisting the change. But time and effort ought to be spent re-optimizing your website for HTTP/2 as a way to see some benefits. Definitely a website optimized for HTTP/1 and simply switched to HTTP/2 with none modifications gained’t do any higher.

Key factors:

  • Usually HTTP/2 is faster when there’s no caching or performance optimizations. The advantages of HTTP/2 are compounded the more severe the web page and the more severe the latency.
  • On a web page that’s already fairly mild, the achieve isn’t so apparent. Optimizing for HTTP/1.1 may be just as efficient as utilizing HTTP/2 .
  • The knowledge of not combining when on HTTP/2 does seem to face true, since in all of the checks where combining was carried out, HTTP/1.1 was the winner.
  • Server optimization, which is out of most of our management, could possibly be an enormous issue. Using Cloudflare might help here.
  • Optimizing specifically for HTTP/2 might yield even higher benefits

As a remaining notice, another angle to think about is that even if the velocity isn’t astoundingly better, HTTP/2 has the good thing about less power consumption.

What have your experiences with HTTP/2 been?