LiteSpeed Hosting as
Cloudways Alternative

  • Select the hosting plan with the LiteSpeed Web server and NVMe SSD storage
  • Choose a shared server geographically close to your visitors
  • Don’t go for the Cloudways hosting services – it is the worst value for money
  • Study the list of standard software suite which is provided in this report
  • Recognise that in the current market, the hosting provider size is its liability
  • Further analysis of the hosting providers can be found in our separate report
  • In the US or Europe, host your website with MechanicWeb – the best value provider
screenshot from Facebook

No Time to Read? No Problem

Here is everything you need to know about Cloudways but were afraid to ask:
  • In the admin area on the Server Management > Vertical Scaling tab, Cloudways states that “the 1 GB and 2 GB plans are for testing purposes and are suitable for staging websites. For production websites, 4 GB (or higher) is required. “
  • Vultr, Linode, and DO advise against using their cheap shared vCPU plans for production websites, especially eCommerce websites.
  • Unlike the rest of the hosting industry, Cloudways does not offer malware scanning or malware protection. Cloudways instructs its customers to sign up for Sucuri’s subsidised subscription for $10 per month; any extra domain will be charged at $16.5 per month.
  • Affiliates know it’s easy to hack unprotected Cloudways accounts. They recommend opening multiple small Cloudways accounts of 1GB each so you do not put all your eggs in one basket.
  • Those who advocate using Cloudways do so by encouraging the use outside of the technical specifications of Cloudways and its vendors. This is a reckless and unprofessional approach.
  • If you choose LiteSpeed Web Enterprise shared hosting, you can handle up to ten times more peak traffic for the same price.
  • There are some shared hosting providers that offer hardware that is 90% faster than Vultr HF.
  • Aside from the fact that Cloudways does not offer email hosting, their technical support is mediocre, and the platform does not offer root access
  • You are not alone in your struggle to understand why anyone should be continuously advised to use cheap shared vCPU plans, misleadingly referred to as VPS!
Executive Summary is 15 minutes
Full article is 55 minutes read
Last modified on September 26th, 2021

Executive Summary

First of all, congratulations on this well-written article. It is very thorough and comprehensive!
I think the most important question customers should ask themselves is how “managed” they want their hosting solution to be. The definition of “managed hosting” varies from company to company. At my company WPAligned, we focus on agency owners. We’re geared for business to business and we use LiteSpeed. Most of our clients do not want to manage the hosting platform themselves.
In my experience, there is no perfect hosting solution, there is only a perfect solution for a particular client.
Congratulations again on this article!

What to expect from this study

Cloudways is often perceived as a provider of advanced hosting services that can offer professionals the best performance. If you read our study, you will see that Cloudways excels in dubious marketing, but its services are not the best. The other hosting providers offer almost twice as fast hardware, but their services are four times less expensive.
LiteSpeed Web Server has fundamentally changed the web hosting industry in the last few years. The LiteSpeed Cache plugin is one of the best site acceleration plugins and is offered on LiteSpeed web hosting for free. LSCache includes a server-level cache and a collection of optimisation features. Today, it is much easier to choose the right web host:
  • First and foremost, they should offer new standard software like CloudLinux CageFS, LiteSpeed, cPanel, Imunify360, ModSecurity, and MailChannels.
    • CXS/ImunifyAV+ is a better malware scanner than Imunify360, but it should only be used by a sysadmin team that is experienced in configuring antimalware software. The addition of the CSF Server Firewall complements nicely ModSecurity, which is a Web Application Firewall.
  • The second aspect to consider is the speed of the hosting provider’s hardware. The spread in the current market is at least three (!) times, so these are not marginal differences.
  • Thirdly, you should make sure that the sysadmin of your hosting provider is able to eliminate various internal delays that slow down the database response.
Benchmarking plugins for WordPress can provide an assessment of hardware speed and sysadmin capabilities. This article discusses several FREE plugins and presents benchmarks for different hosting providers. Our review explains how to interpret the benchmark results. You can use our performance tables to quickly evaluate your current web hosting provider.
Our benchmark study shows that with the help of a couple of plugins under WordPress, it is possible to identify the advanced hardware used by the hosting provider. For example, it is possible to determine without a doubt which Intel Xeon servers such as E3-1270 v6, E-2136, E-2288G, or E-2388G are being used. It is also possible to identify and clearly distinguish between AMD EPYC Zen 2, EPYC Zen 3, Ryzen Zen 3 and Ryzen Zen 2/ Zen+ servers.
The AMD EPYC Zen 2 is comparable to best servers from Google Compute Engine, DigitalOcean, Linode and Vulr HF. Other hardware listed above easily outperforms cloud computing services. The AMD Ryzen Zen 3 is 90% faster than the most popular platforms on Cloudways.

Cloudways is deceiving its customers

Cloudways provides SHARED hosting that is way overpriced and offers restricted hardware resources. In general, most hosting providers offer both cheap shared hosting plans, inexpensive VPS with shared CPU resources,  and expensive VPS hosting plans with dedicated CPU resources. This is also true for Vultr, DO and Linode. Their shared vCPU plans are affordable and starting at $5 per month. They also offer dedicated vCPU plans starting at $30 at Linode, from $40 at DO, $60 at Vultr.
Shared vCPU plans were initially available only to professional developers testing new web applications that were not yet in production. All vendors recommend limiting such plans to low-traffic sites. Developers who require root access will be fine with using shared VPS plans. Cloudways does not permit root access on their platform, however. Cloudways offers VPS accounts with shared resources and without root access. It is a shared hosting plan disguised as something else.
Cloudways hides information about the shared nature of hardware resources.  However, you can find it in its management panel under the Server Management > Vertical Scaling tab. It states that the 1 GB and 2 GB plans are for testing purposes and are suitable for staging websites. Production sites must be served by 4 GB plans ($50 per month) or higher. You can access Cloudways’ management panel only after you sign up for one of their plans. You will have a hard time finding a similar wording on the Cloudways website!
Speed and TTFB is not the only thing you should care about. Customer service, security, reliability, price and features are also important factors.
I have tried Cloudways in the past, and the WP performance was abysmal and the support was non existent.
People think that Cloudways provide hosting, but they do not. They are simply reselling other companies’ unmanaged shared hosting and slapping a management interface on top of it. That seems to be pretty much all they do.
The incorrect wording used on Cloudways’ website when describing service offerings leads users to believe they are signing up for dedicated CPU VPS, when in fact, they are signing up for shared vCPU plans.
One of our readers, Erik Klok, contacted Cloudways sales to verify if the above claim was valid. The shocking lie he received in response makes even the most cynical used-car salesman blush:
“We provide dedicated servers that come with neat IP that has not been used in any shared environment. The droplet is dedicated. The disk is only for them, the CPU is for this Droplet, we generate specific new machines and we launch all resources from zero.”
The reader of our article should open and read the definition of a dedicated Droplet provided by DigitalOcean. All Droplets offered by Cloudways are called basic Droplets. With a basic Droplet, you get a shared virtual CPU. A dedicated Droplet with dedicated CPU resources is available starting at $40 per month. Dedicated plans on Vultr also offer dedicated CPUs, but they start at $60 per month.
Here is an analogy. Consider DigitalOcean, Linode, and Vultr as the best-known brands for advanced gaming laptops with dedicated graphics cards. At some point, they all decided to release a cheap version of the office laptop – without a dedicated graphics card. Shady companies resell cheap office laptops but call them “gaming laptops” and charge much more than the office laptops are usually worth. Social media is full of exciting reviews from affiliates about how fast they can run a Solitaire game on these special edition “gaming laptops”. Many people buy these computers after reading such reviews.
Are you actually getting ‘dedicated’ server resources with CLOUDWAYS or is it yet another HOSTING SCAM?
In the hosting industry, there is quite a con going on with ‘dedicated’ VPS resources. The general consensus is that a VPS is always better than shared hosting because you supposedly have dedicated CPU and dedicated RAM. Anyone repeating this naive nonsense to you is frankly ignorant of the technical and financial realities of Web hosting!

Cloudways compromises security

Cloudways is misrepresenting its shared vCPU plans, but that’s not the only problem. In contrast to the other hosting providers listed in this article, Cloudways does not provide default protection against malware. 94% of hacked websites are powered by WordPress. The compromised security of Cloudways plans is not known to most users who go along with Cloudways’ marketing hype. This is a red flag for professionals, who know that WordPress security plugins are no good – proper protection against malware and ransomware can only be achieved at the server level.
GDPR compliance depends on how well you protect EU customers’ personal data from ransomware attacks. Cloudways offers paid Sucuri subscriptions for $10 per month for the first domain; all other domains will cost $16.5 per month per domain. If you do not find this offer, which is well hidden on Cloudways’ website, you could be accused of not taking personal data protection seriously. Failure to comply with the GDPR can result in a fine of up to 4% of your annual turnover.
Ignorant Cloudways Affiliate scammers often recommend to rely only on regularly updating all plugins and WordPress core files. Let us use an anlogy. Do you agree that the lack of antivirus software on your PC can be compensated for by automatically updating the Windows core files? Most people will say that you need to use both.
The most sophisticated affiliates recommend the free Wordfence as the only malware protection. Wordfence’s mission-critical updates are released to free users 30 days after they are released to premium users. Also, free users do not have access to the real-time IP blacklisting feature. For personal websites, free Wordfence is a compromised choice. However, it does not meet your needs if you are hosting a business website or an eCommerce website. The paid version of Wordfence costs virtually the same as the discounted version of Sucuri offered by Cloudways.
An analogy can be drawn here. You need a bank vault. Most banks hire a security company (Imunify360 / ModSecurity) to fix the vulnerable firmware in the electronic locks of the safe deposit boxes as soon as problems are discovered. One new dodgy bank (Cloudways) does not offer this service. The fine print offers its customers subsidised but still costly direct services from another security company (Sucuri). Affiliate bank vault experts strongly advise you to use the excellent services provided by the dodgy bank. They also recommend that you subscribe to the free promotional assistance of the latest security company (Wordfence), which will fix your electronic locks 30 days after a lock security breach is announced. What words would you use to describe the affiliate  expert’s professional background?
Through HackRepair.com, I have provided malware and hack repair services to quite a few Cloudways customers over the past few years. I have always wondered why Cloudways is so “popular”. Their creative affiliate marketing is not new, but their services are not modern either.
You should be a tech-savvy person who can handle primitive control panels and email-only non-responsive customer service. Cloudways’ $10 per month plan provides you with 0.5GB RAM to serve a simple website with above-average storage requirements.
For a similar monthly fee, you can easily find a LiteSpeed / cPanel shared hosting service that is more suitable for small businesses. Such services are much easier to use and less costly in the long run, leaving HackRepair.com with fewer customers.
Hosting companies will go the extra mile to solve general WordPress-related problems, and most have satisfied customers with many positive online reviews. Prices for shared hosting services range from $10 to $20 per month and offer a good level of technical resources for high-traffic websites. If you opt for it, you’ll enjoy LiteSpeed Cache with HTTP/3, which comes in handy for optimising the speed of your website.

LiteSpeed is cutting-edge technology

LiteSpeed has created a new ecosystem with its enterprise software LiteSpeed Web Server, LiteSpeed Cache for WordPress, and QUIC.cloud CDN for the fast delivery of assets, including HTML, to end-users. You will get at least 10% or more efficiency gain on busy servers with LiteSpeed Web Server; LiteSpeed Cache is a better free alternative to paid WP Rocket for optimising your site’s page speed rating; and QUIC.cloud CDN offers the best, most sophisticated HTML edge delivery available.
LS cache is the real deal! The fastest growing cache plugin right now. I think it’s so incredible that a webserver company decided to write their own cache plugin and release it to the public for free. They’re probably more qualified than anyone else as WordPress caching should be done at the server-level before code-level. For many people (beginners and pros alike), LS cache is the gold standard…their holy savior in the world of cache plugins. LS cache (when used with LS server) has the advantage of Varnish and could outperform any other caching. LS cache’s advanced settings can really shine if you know how to use them, especially private cache and ESI. That’s some cutting-edge stuff. I can’t believe they would do that for free!
LiteSpeed Cache is a full-page cache for a website’s dynamically-generated pages and a site-optimization plugin. This article does not address the modern battle of caching plugins to speed up the rendering of your website in the browser. Our separate, very popular study describes the best techniques implemented so far by various caching plugins to improve your Core Web Vitals score. Feel free to take the time to study our research to appreciate the value that the completely free LiteSpeed Cache plugin offers. There is no better option for web designers among a multitude of other paid plugins!
I’ve been using NameHero in the US and Krystal in the UK for a few years now. Both use LiteSpeed Enterprise. My clients’ sites are much faster than with the previous hosting, which used Nginx. I was able to get A grades on GTmetrix, and the sites feel noticeably faster. LiteSpeed and LSCache are regularly updated by the developers, which further improves performance. I wouldn’t go back to Nginx. LiteSpeed is the way to go for sure. LSCache 4.4 came out the other day, which offers more optimisations to increase speed.
LiteSpeed Cache on WordPress can be used to its fullest only if your hosting provider installs the LiteSpeed Web Server Enterprise software. If you want to take full advantage of LiteSpeed, you should move your website to an appropriate hosting provider.
I have hosting with LiteSpeed and Cloudways so I can write about my experience. I am a web designer working with clients from different countries. If I do not know any good local LiteSpeed hosting companies, I choose Cloudways. My experience shows that LiteSpeed is far superior to Cloudways. Oxygen Builder works better with LiteSpeed Cache. To compare both hosting options, I have a specific example. I had an Oxygen website on Cloudways with a PSI speed of 75 after all optimizations using WP Rocket. The client was unhappy. I have built many websites with the same toolset hosted on LiteSpeed servers using LiteSpeed cache. Each time I have achieved a score of more than 90 with the PSI test. So I am pretty sure LiteSpeed servers with LiteSpeed cache are the better solution! If I have the opportunity to work on LiteSpeed servers, I will never use Cloudways.

A workstation grade server runs twice as fast

Many small hosting companies have adopted this new ecosystem. Some enthusiastic companies offer LiteSpeed along with the most advanced server hardware currently available. Serwer.io offers workstation-grade AMD Ryzen Threadripper 2950X processors with turbo speed up to 4.4 GHz and ECC DDR4-2933 RAM. MechanicWeb offers a workstation-grade Xeon W-1290P processor with a 5.3 GHz Turbo and DDR4-3200 ECC memory, Intel Xeon E-2288G with a 5 GHz Turbo and DDR4-2666 ECC memory, and Zen 3 AMD Ryzen 9 5950x with turbo speed up to 4.4 GHz and ECC DDR4-3200 RAM. Both companies maintain highly optimised software settings on their servers.
A server-grade system is not good without ECC memory. A corrupted database record is not something you want in an eCommerce environment. That is why professional data centers offer ECC memory on general workstation-grade servers wih Ryzen CPU. Systems without ECC memory are only suitable for gaming servers.
Hosting plans from Serwer.io and MechanicWeb outperform Vultr HF 4GB on Cloudways in every single CPU Core benchmark measured by benchmark measurement plugins like WPPerformanceTester and WPBenchmark. Overall, their speed is much higher than the best Vultr HF plans on Cloudways. Serwer.io offers 32% faster speed than Vultr HF, and MechanicWeb offers 46% faster speed on Intel Xeon servers and 92% faster speed on Ryzen Zen 3. According to the Performance vs. Price chart, both companies offer 10 times better value than Cloudways. Serwer.io offers five times more traffic for half the price, while MechanicWeb offers nearly three times more traffic for a quarter of the cost of Cloudways.
LiteSpeed_vs Cloud
In the graph, the performance value represents the number of visitors per second that can be served by the plan with a fresh installation of WordPress, Elementor and Hello Theme. We remove all caching layers. Cloudways does not allow OPcache to be disabled, so their values measured for this test are higher than they should be for a fully comparable test. To reduce the efficiency of OPcache, we reduced its memory limit to the minimum allowed value of 9 MB. Traffic is measured using K6 load test.
Another more readily available choice offered by various Data Centers is the Xeon E-2136  (All-incl) with 6 CPU cores, 4.5GHz Turbo and DDR4-2666 ECC memory or about 10% slower E3-1270 v6 (EthernetServers) with 4 CPU cores, 4.2 GHz Turbo and DDR4-2400 ECC RAM. These servers are  better than the best instances of Vultr HF 2 GB servers on Cloudways, but you can find deals that are six times cheaper than 2 GB Cloudways plans at companies like EthernetServers or HostMantis.  
The performance of the Xeon E-2136 CPU is 10% better than that of the Xeon E3-1270 v6. Yet data operation has improved by nearly 2.5 times between these two models, which were introduced just a year apart. The Intel Xeon E-2288G was introduced a year later. It offered a similar 10% improvement in CPU speed over the Intel Xeon E-2136. This did not result in any further significant gain in the data processing.

AMD EPYC servers are not as fast as they claim

AMD EPYC servers improve CPU density per server, resulting in reduced software licensing costs and data centre space rent per end-user. Estimated savings of about 30% per visitor request are reflected in cheaper hosting plans. With CPU core frequencies around 3 GHz, AMD EPYC hosting is less powerful than the options shown above but is very similar to Vultr HF and DO Premium Intel servers offered by Cloudways. With a reduced cost of ownership, AMD EPYC is best suited to scaling a hosting business. For a good example of a hosting provider using these servers, consider NameHero using AMD EPYC 7302. Their tri-annular Turbo Cloud plan costs six times less but offers slightly better performance than 4 GB Cloudways.
AMD EPYC Milan servers will offer up to 32% more performance than AMD EPYC 7302 Rome servers used by NameHero. In the next few years, the AMD EPYC 7343 will be the optimal processor for the large scale servers. There is quite a shortage of new AMD EPYC processors due to their higher performance. While NameHero has confirmed their ability to find a few to upgrade their offerings, most web hosting companies will have difficulty upgrading to the newer AMD EPYC processor. You should check with your hosting provider to determine the exact model of a processor included in your plan, as AMD EPYC can imply an older generation with low performance.
Why do we use hosting servers? The answer is simple: to host our websites. But there are so many hosting service providers, how do you choose one? Well, for that, we need to consider some key things like: 1- Performance, 2- Pricing, and 3- Security.
I would say that this article is a gem, an eye-opener (even for professionals) and a true friend when it comes to choosing a hosting server. The research is rich in valuable information, but at the same time, it’s free. I want to give a big shout out to the people behind this research. Hats off to you guys. I can not thank you guys enough for providing us with so much helpful information.
It is interesting to compare the performance of NameHero’s current EPYC Zen 2 server with that of Nexcess’ AMD EPYC Zen 3 server (see the second table below). Nexcess is part of Liquid Web, and NameHero uses Nexcess’ servers to serve its own customers. A very similar setting for the high availability cloud server deployment is used for both companies. In both the Database section of the WP Benchmarking plugin and the WP Performance section of the WPPerformanceTester plugin, the results for their servers are nearly identical and rather dull. NameHero will soon be offering the EPYC Zen 3 servers used by Nexcess. This is not a game-changer as they provide similar performance. However, math calculations improve almost threefold when you switch from EPYC Zen 2 to EPYC Zen 3 servers. Compare NameHero’s and A2Hosting’s (EPYC Zen 2) with BKHosting’s  and Nexcess’s (EPYC Zen 3) data in the tables below to verify it.
During the last six years, NameHero has been growing fast and now serves more than 750,000 websites. In addition to its reliance on LiquidWeb servers, NameHero is planning to open a data centre in Kansas City. We expect the new EPYC Zen 3 servers to be used there.

The new AMD Ryzen Zen 3 is truly incredible

The AMD EPYC platform, contrary to popular beliefs on the web, is still relatively slow.  You should be aware of the big performance difference between AMD EPYC and AMD Ryzen servers. If you are looking for a fast server response time, watch out for the long TTFB response time in our tests for NameHero and BKHosting (please refer to the table below). For the same uncached site, MechanicWeb’s Ryzen Zen 3 server showed three times (!) lower TTFB. Your visitors will certainly benefit from a more responsive website if you choose a hosting provider with AMD Ryzen Zen 3 servers.
Benchmarks show that the Ryzen Zen 3 servers are on average twice as fast as EPYC Zen 2 servers. MechanicWeb uses a Ryzen 9 5950x with 16 CPU cores as the optimal AMD Ryzen Zen 3 server. An expensive 5980X “Chagall” Ryzen Threadripper Zen 3 server with 32 cores will require much more demanding cooling. It is unlikely to be popular with speed-junkie hosting providers. AMD Ryzen Zen+ (Serwer.io) and AMD Ryzen Zen 2 (Nexus Bytes) have almost identical speed metrics to Intel Xeon E-2136 (All-incl). The move to AMD Ryzen Zen 3 (Mechanic Web) shows a big jump of about 45% in server performance. AMD has taken a big step forward! As a result, it can outperform by 30% the best Intel workstation servers like the E-2288G used by MechanicWeb.
With the introduction of the Xeon W-1370P and Xeon E-2388G processors with 8 CPU cores, Intel might be trying to catch up to the speed of AMD’s Ryzen 9 5950X. Once MechanicWeb or another vendor gets their hands on the new ‘Rocket Lake’ workstation-grade Intel CPUs released in Q2 and Q3 2021, we will update our benchmarks.
We believe that the new Intel Rocket Lake processors will move slightly closer to Ryzen Zen 3 speed but will remain about 15% slower. Regardless of this, the price-performance ratio will be unfavourable due to the two times lower CPU count. A healthy degree of scepticism about Intel capabilities to impress is caused by the zero performance improvement introduced between Intel Xeon W-1290P (workstation grade, Comet Lake) and its predecessor Intel Xeon E-2288G (server grade, Coffee Lake). Our benchmarks measured with MechanicWeb’s Xeon W-1290P servers in Dallas were identical to the ones measured with the MechanicWeb’s Xeon E-2288G servers in Los Angeles.

Benchmark Tables: a first look

Top_LiteSpeed_providers
In the Benchmark Study section of this report, we describe our methodology for performing five different tests, which are summarised in the table. Companies such as Nexus Bytes, EthernetServers, and NameHero appear to offer servers with unoptimised MySQL settings. It is likely caused by the high availability cloud shared server deployment resulting in poor performance scores in the “Database” and “WordPress Performance” sections.
Top Nginx providers
Cloudways’ 4 GB Vultr HF plan performs poorly in all CPU performance tests (greenish colour highlighted). The vast majority of LiteSpeed vendors in the first table outperform Cloudways on CPU processing speed and price. Of course, you can find other overrated providers with even slower service. The second table contains some examples. The fact that WP Engine seems to be using old Google servers is a surprise. The Raidboxes account performed horribly in the K6 load test. It is rather puzzling to see Kevin Ohashi mentioning Raidboxes as a top recommendation for shared hosting in his latest review.
Second_Tier LiteSpeed_providers
The third table extends the benchmarks to other LiteSpeed hosting providers. Most of them are not worth your time. For example, WPX’s $ 21-month plan is too slow and will serve 2x fewer visitors than MechanicWeb’s $3-month Smart plan. The reason why affiliate scammers like Matthew Woodward claim WPX is the fastest WordPress hosting solution is puzzling.
Closte is 15% faster than Siteground and 10% faster than Kinsta. WPEngine is 38% slower than Closte. The Google Compute Engine platform is used by all four companies. This is a big surprise as all four companies were tested from the same London data centre. They probably use different subscription plans, but such a big difference is unexpected.
Most hosting providers in this third table severely limit the disk speed for your package. Hostinger’s speed of 1 MB/s is not only a violation of common sense, but it shows incompetence. The WP Benchmark plugin could not even run Filesystem test on Hostinger’s account. If your hosting provider is not able to run the performance test plugin, why would you choose such a crappy server? Do not believe Darell Wilson when he says Hostinger is the fastest and best option. Check out our benchmarks to see the depth of the ethical swamp that affiliate programs create!
I am a big fan of LiteSpeed and its QUIC.cloud CDN. However, I am cautious. GDPR is a pain for European web developers, and it’s especially problematic in Germany, where no data protection officer will agree to using a CDN. The ePrivacy regulation requires that any transfer outside the EU, including IP addresses, must be approved. Several law firms advised me against using a CDN or set aside money for litigation.
For me, a test of the European hosting provider being only “uncached” is not enough. It’s essential to benchmark the delivery of a significant amount of cached content.
Raidboxes have a custom Nginx configuration and modify WordPress core files. Their CGI cache can handle many visitors, even on the cheapest plan. However, you cannot take advantage of WP Rocket or LSCache’s using such hosting.
It is interesting to benchmark European LiteSpeed providers with Raidboxes to compare their cached content delivery.
Thanks  for the extra tests. I read through both of your articles on hosting and CDN. Great articles! Now I understand why I recognized your site: the article about NitroPack is pure gold. You write great articles that are well researched and in-depth!
Andre Baum
Andre Baum
Andre, thank you so much for your comment! We have taken your suggestion and tested the MechanicWeb servers in the Hetzner data centre in Germany against Raidboxes. The main tests are performed with loader.io for fully cached webpages and K6 load testing for uncached webpages. With fully cached content, MechanicWeb with LSCache is 1.5 times faster and can handle significantly more traffic. We stopped our tests at the rate of 2,000 visitors per second, which is nowhere near the limit of the LiteSpeed server. Even with such unrealistically high traffic to the webpage, the TTFB remains under 60ms. The Raidboxes performed eight times worse than MechanicWeb with LiteSpeed Enterprise Web Server in the performance test without cache. Considering that MechanicWeb is about six times cheaper, we recommend trying it out! It will work perfectly well without CDN.

What's the best way to choose a web host?

All of this would lead to professional communities being more interested in learning what’s going on in the LiteSpeed ecosystem and discussing the pros and cons of the various LiteSpeed hosting providers and hardware they are using on their servers. After all, there are already 5,000,000 sites using LiteSpeed. However, that’s a far cry from what you read on Facebook forums or watch on YouTube these days. The purpose of this study is to fill the knowledge gap.
I am really amazed at how detailed the article is. I learned quite a lot! Kudos.
The pandemic has made clients more hands-on when it comes to creating a website. Unfortunately, they chose hosting because of advertising and marketing, and I agree that today the most hyped products have the highest number of affiliates. However, I think it is our job to fix this problem.
I appreciate the way your article educates and guides web designers, especially with the vast abundance of hosting providers out there today.
In the future please let me know if there any other articles like this. I really learned a lot!
Marion Paolo Abagar
Marion Paolo Abagar
The rest of the article is divided into four relatively independent chapters:
  • Chapter 1. We give a detailed overview of the standard software options offered by all reasonable LightSpeed web hosting providers. Before you choose your hosting, you should read this chapter!
  • Chapter 2. You should read the detailed chapter explaining the cheating and fraudulent schemes of Cloudways and its affiliate community. We strongly recommend you to study it if you are still angry at us after reading our direct insults against Cloudways and their iconic services.
  • Chapter 3. In the third chapter, we explain the technical details of our benchmarking study. Read it if you want to quickly test the actual quality of hardware resources provided by your hosting provider. You’ll also get an objective look at their sysadmin team – can they fine-tune software settings to optimize database performance?
  • Chapter 4. The final chapter covers advanced hosting options for extremely busy websites. You should study this chapter if you have more than 100,000 visitors per month. The new generation of semi-dedicated shared hosting can easily handle traffic of up to 3,000,000 visitors per month without requiring you to switch to VPS hosting.
By contacting us, readers can actively contribute to our study. You can read the opinions of about 60 people on this article, and feel free to share your own opinions with future readers.
I went through the article and all the comments that all knowledgeable folks have left. I do see some familiar faces, but beyond that, I see familiar thoughts. Most of what I’ve had in my bag of experiences is already shared by other people, like Bobby and Andre, so I doubt I’ll be able to add any extra value to the article. Having said that, I appreciate that it is a very well-written and helpful article that I hope helps as many WordPress users as possible, especially those who are new to the game. Thank you for writing it, and keep churning helpful articles like this one!
It is a rare option to read a study that is not influenced by affiliate fees! Read the rest of the article to learn more about modern web hosting.

Introduction

In August 2021, Google has updated its algorithms to include Page Experience metrics based on Google’s Core Web Vitals. It was a significant change that has an impact on ranking. Things like First Contentful Paint (FCP), Largest Contentful Paint (LCP), Flashes of Unstyled Content (FOUC), Cumulative Layout Shift (CLS) are creating a nightmare of optimisation concerns for web designers. This new ranking signal results in a sudden and urgent shift towards page load speed optimisation, lightweight HTML, CSS, and reduced JS payloads.
But the actual picture is more complicated – the page loading time of many websites is pushed further up because of the wrong choice of the hosting providers. The classical cycle of blame, the Hosting provider – Caching plugins – Web page building tool – Web designer – Hosting provider, requires optimisation of each part of the chain.
This post provides some ideas for optimising your website’s performance by correctly choosing your hosting provider and associated infrastructure.

Chapter 1. Hardware/Software requirements

Standard offering

A modern shared web hosting provider must offer at least the following:

  • LiteSpeed Web Server –  Web Host Professional version on AMD EPYC server; Web Host Essential is enough for the Xeon or Ryzen server with a smaller number of CPU cores
  • NVMe Storage – from 5 GB per domain
  • Disk I/O speed quota of more than 30 Mb/s; the best value is 50 Mb/s. Do not compromise on the disk speed. Ignore offers below 50 MB/s if you can.
  • 1 GB RAM or more; DDR4-2400 ECC or higher; the best value is DDR4-3200 ECC
  • 1 GB RAM per 1 CPU core is the sweet spot. You might want even more RAM for a bustling eCommerce website as each parallel PHP request will require RAM resources
  • 1 CPU core at 3.2GHz+ frequency for AMD EPYC core; aim for 4.0 Gz+ for Intel CPU cores; 1 CPU core per 20 entry processes is a golden standard
  • The recent AMD EPYC Zen 3 CPUs at 3GHz+ provide better value for money compared to Intel’s Xeon; AMD Ryzen Zen 3 workstation-grade servers offer the best results, there is a good chance that they are three times faster than your current server; please double check that you will have ECC RAM
  • AMD EPYC Zen 2 shows terrible performance in operations with data and string manipulation; AMD EPYC Zen 3 has largely fixed these issues
  • Don’t take the hosting plan if the company cannot list the specific version of the microprocessor and the hosting server’s memory speed. Don’t trust generic words about fast servers; it means eight years old server with DDR3 1600 MHz RAM at best
  • Simultaneous entry processes – aim for 16 per busy website or 4 for a simple website
  • CloudLinux with CageFS – to uniquely encapsulate each customer on the shared server
  • cPanel access
  • Automatic daily remote offsite backups using JetBackup
  • Lets Encrypt – free SSL
  • Softaculous WordPress installer/staging environment
  • Capability to run Cron Jobs using a command-line program such as curl
  • Persistent object cache like Redis or Memcached Server with SASL for the eCommerce website
  • Malware scan and web application firewall (WAF) – Imunify360/ ModSecurity is the industry norm for cPanel accounts; you might also find even better option like CXS/ImunifyAV+ with added server level CSF WAF
  • MailChannels SMTP relay service with IP blacklist protection – for improved email deliverability; MailChannels is using spam filter to stop your naughty neighbours before the origin IP gets blacklisted
  • HTTP/3 service enablement; only LiteSpeed Web Server supports production-ready HTTP/3; this will give you a further boost when using QUIC.cloud CDN
  • Free migration support from another provider
  • Internet network uplink connection of at least 1 Gbps via renting a server from a reputable Data Center integrated into global Internet Exchange Points (IXPs)
  • Cloudflare Railgun integration was important in the past, but its benefit is not requird on LiteSpeed servers, which caches the dynamic part of HTML and stores it as static files. *
* A new update. As of June 2021, QUIC.cloud CDN is in production stage, with the cache lifetime extended from several hours to a day. The Edge delivery of HTML files via QUIC.cloud is tightly integrated with the LiteSpeed caching plugin even for dynamic pages. You should use QUIC.cloud instead of Cloudflare CDN.

LiteSpeed and NVMe SSD combo

You have to consider the hosting configuration based on LiteSpeed WebServer software and SSD disk with a high-speed PCIe socket for Non-Volatile Memory Express interface protocol (NVMe).  Such a winning formula boosts about four times the speed of specific internal processes on a server, removing the bottleneck in its performance. Servers with such improved technical specifications will carry heavy traffic with more internal resources and degrade their response time more gracefully. It is not just empty words – sensible shared hosting providers increase fourfold the number of simultaneous entry processes and the disk I/O speed quota when introducing this improved configuration to their offerings.
Servers hardware features
NVMe SSD looks like a RAM, not a disk
LiteSpeed Web Enterprise provides additional WordPress Brute Force Attack Protection. A “brute force” login attack is an attack on a website that attempts to gain access to the site by repeatedly guessing the username and password. WordPress is the most popular CMS and, therefore, a frequent target of this type of attack. LiteSpeed protects shared hosting WordPress environments from large-scale brute force attacks that have the potential to bring down entire servers.
LiteSpeed appriasal

CPU speed and entry processes

Please be aware that any request in WordPress is single-threaded and uses only a single CPU core. The faster the thread, the quicker the PHP request or MySQL query to the database will run. High-frequency CPUs provide the highest single-threaded performance. However, benchmark tests shows that AMD Ryzen Zen 3 will outperform Intel CPU with the same clock speed by as much as 30%.
If your website requires a high amount of server resources and CPU processing power, you first have to check the number of allocated simultaneous entry processes on your plan. It is as critical as the number of available CPU cores. Of course, it is useful to have more than one CPU core for the busy website with simultaneous concurrent requests coming from different visitors. It is unlikely your case unless you have more than 50,000 clicks per month. Even then, the higher the clock speed of the CPU, the faster your page will be generated.
You will also see increased RAM allocation with the increased number of available CPU cores on a good hosting plan. It is common to have a similar number of GBs of RAM and the number of allocated CPU cores (such as 2GB of RAM and 2 CPU cores). Adding more RAM per CPU will help to serve multiple simultaneous PHP requests. If you keep many domains under your account, then having more RAM could be beneficial. In this case, more RAM is needed to store all databases of different websites.
OpenLiteSpeed
The number of allowed simultaneous entry processes is the crucial parameter determining the speed of processing the uncashed requests.  An entry process allowance is the number of visitors you can serve at the same time when someone is accessing your website. If a user accesses a page of your website, that is a single-entry process. When this exceeds the maximum number of the entry process, you will likely receive an Error 503 when a page is accessed. It usually happens if a long queue is created when many users start accessing your website’s pages simultaneously.
It would be best to have about 16 entry processes per crowded eCommerce website. Your server will cope perfectly fine with delivering cached webpages using just four entry processes. The rule of thumb is to have about 1 CPU core per 20 entry processes, roughly matching the number of threads per core in most server processors. As a result, you have to watch that the number of CPU cores is scaled up sufficiently to reflect the entry processes’ allowance. A modern single logical CPU core on the LiteSpeed Web server should be capable of enduring a continuous load of up to ten HTTP requests per second.
Kinsta hosting corporation, the darling of so many WordPress bloggers, provides four entry processes per website (they call them PHP workers) on their Business 1 plan, which cost $100 per month and is restricted to 5 domains. As you will see in our separate study, $20 per month will buy you a service with 100 entry processes. To have 16 entry processes per website, you should upgrade to the $1,500 per month Enterprise 4 plan on Kinsta.

LiteSpeed Cache Engine

LiteSpeed Cache Engine is a built-in feature of LiteSpeed servers; it provides powerful acceleration of dynamic content. An LSCache plugin submits web application rules to LiteSpeed Cache Engine. Cached files are not stored in the WordPress file structure, and the plugin does not execute any of the caching tasks.
LiteSpeed Cache works in a similar way to Varnish. It implements server-side caching. Dynamic content that is not limited to PHP pages is cached by LiteSpeed. Unlike Varnish, LiteSpeed has integrated its Cache Engine into its Web Server, eliminating a layer of reverse proxy. This results in higher efficiency for static content.
If needed, LiteSpeed can store pages in a public cache, a private cache, or it can leave a page uncached. Simple websites often work with public caching only, offering the same cached page to every visitor. If a specific version of a page can be shown to a specific user multiple times as a static page, that page is cached privately. A shopping cart is a classic example of data that should be privately cached. If the response is different each time the page is requested (and therefore cannot be served as a static page to anyone), the page should not be cached.
By combining public and private content on one page, ESI (Edge Side Includes) markup lets you manage the page cache dynamically. Typically, you start with a public page and then add “punched holes” for private or non-cached content. Each element of the page can be handled separately, and you can cache it in the public cache, in the private cache, or not cache it at all.
ESI means that you cache most of the page publicly and add dynamically generated specific content that is different every time the page loads. Most of the HTML, inline scripts, images, etc. remain static, even if several different users access a user-specific page.
An example of this would be a shopping page with a shopping cart widget. The content of the widget needs to be cached privately because it differs from customer to customer, but the rest of the shopping page is the same for everyone. In this case, you would cache the page publicly, while using ESI to cache only the widget privately. The mixed content would be reassembled and delivered together. Using ESI, LiteSpeed Cache Engine can fully serve dynamic web pages from cache without the need for backend server intervention. This way you can serve more eCommerce visitors without increasing CPU resources on your hosting plan.
Edge Side Includes in LiteSpeed
Edge Side Includes in LiteSpeed
By combining HTTPS/HTTP2 termination, caching, and web server into a single tool, LiteSpeed offers a higher level of simplicity and ease of maintenance compared to Varnish. Some advanced paid caching plugins can also use ESI markup, but they need to call backend PHP scripts to populate dynamic content.  As described in the following section, such calls can be accelerated using Redis or Memcached in combination with OPcache. However, they cannot match the efficiency of LiteSpeed’s server-side caching.
The loading of  hybrid ESI dynamic pages is now greatly improved.  LiteSpeed Cache 4.0 introduces Guest Mode for WordPress. The site will load quickly on a user’s first visit to improve the Core Web Vitals Score for your complex eCommerce site. Guest Mode allows you to defer decision making for user-specific content and immediately serve a default cached version. For this first visit only, there are no caching variations and no ESI. You do not have to worry about whether the visitor is logged into your site or not.
Such a page loads quickly because the system does not have to spend time processing ESI. It simply finds the “public” page in the cache and serves it. During the HTML loading process, an Ajax call is made. This is the point at which any factors that were ignored are processed and the “correct” version of the page is loaded with ESI for that user.
The LiteSpeed Cache plugin is designed to understand the rules of the app and implement public, private and ESI caching as needed. Hosting providers should enable ESI on their server, but most will do it automatically.
Full automatic ESI integration with WooCommerce is one of the benefits of using LiteSpeed servers. The WooCommerce plugin itself specifies which part of the page is suitable for processing ESI markup, and LiteSpeed takes this into account when generating the page. Most caching plugins for WordPress do not include ESI. With LiteSpeed, you get advanced ESI caching for free, making LiteSpeed webhosts the ideal platform for hosting WooCommerce.
Edge-Side-Includes is an XML markup language that was developed to be inserted into HTML to instruct the CDN to combine static and dynamic content. With all other caching plugins like WP Rocket, ESI must be implemented in the CDN. For Cloudflare CDN to work, you need a web host that supports Railgun. Railgun performs delta compression on requests received from a web server. If the page has ESI enabled, the Railgun listener sends only the dynamic difference. The CDN caches the public portion of the page and retrieves the rest from the destination defined in the Edge-Side-Include block. To send the final web page to the visitor, you will have to use Cloudflare Workers to combine both the static and dynamic parts. Using Workers, your software developer must create bespoke custom logic right within the Edge.
Cloudflare Railgun is not yet compatible with RedHat 8, CentOS 8, AlamLinux 8, and CloudLinux 8 because Cloudflare has not updated its software. As a result, many providers that offer Railgun will not be able to use it. You are not alone if this story sounds more complicated than switching to LiteSpeed Web Hosting.
I’m slowly migrating a few servers to OpenLiteSpeed and am happy with the results I see. Lower memory usage and still happy customers. I don’t see any direct performance gains, but when a few sites get A LOT of traffic, there’s no stress anymore. One of the biggest reasons for switching to OpenLiteSpeed is the lower server load during peak times. This alone allows us to run a few more sites on one server, which positively impacts our bottom line.
PS. Saved your site, you have some nice write ups!

Object cache

Many web designers are not even aware of the role of ESI (Edge Side Includes) in delivering dynamic content. By default, ESI is turned off in LiteSpeed Cache, and only knowledgeable users will implement it on their websites. If you use ESI, you do not have to worry about object caching. However, there may be cases where the ESI approach fails, and you need to revert to the old-fashioned Nginx-era approach, such as using an Object Cache.
Object cache such as Memcached, Redis, or OPcache can be activated, but it is not touched if you use a full page cache such as LSCache with ESI. Hosting providers use LiteSpeed Web Server Enterprise, but smaller private VPS servers often use free OpenLiteSpeed software. Note that it does not provide ESI support. If you are not using ESI with LS Cache or using another caching plugin like WP Rocket, you may need Memcached or Redis, and OPcache.
Most cache plugins are automatically considering cart and checkout pages uncacheable. Pages with such private content are dynamically generated each time and served uncached.
For eCommerce WordPress sites with many requests for dynamic HTML pages, adding a persistent object cache like Redis in front of your MySQL database can boost performance and reduce entry processes load. Without a persistent object cache, MySQL database queries will be executed for each request even if the result is identical to a previous query. Redis stores the results of database queries in the RAM, which allows PHP to grab the results of queries that have already been executed. This object caching method allows PHP workers to conserve CPU resources and spend less time fulfilling a request because it removes the need for repetitive database queries.
Some providers listed in our report allow to use the Redis for free but heroic Kinsta charges $100 per month per website for such a privilege. It is not straightforward to offer Redis services on the cloud platforms in which RAM is allocated only when requested by the new process. Kinsta needs to run one Redis container per user account and set up a cluster of these containers. Providing this service requires detailed knowledge of Redis and the allocation of dedicated cloud resources. It is also based on using relatively costly Redis Enterprise Cloud software.
Redis is open-source software that offers various security improvements to its latest versions, including reduced attack surface for hackers, encrypted communication channels, and permissions management. However, by its nature, open-source software like Redis can be configured to use all or none of the available security features. When it comes to using Redis in shared servers, providers should review their security settings and add additional mechanisms.
Be careful with your decision to use Redis on shared hosting. Unless you are positive that your hosting provider has a skilled and highly professional sys-admin team, please don’t go for it. And don’t simply trust a small host provider. There is a high chance that their Redis configuration will lead to an insecure environment in which different users will be able to read and rewrite each other’s Redis memory. Separate Docker containers for separate customers will reduce security risks in shared hosting. However, dedicated Redis instances for individual users in a shared hosting environment is rarely done. Server-wide Redis can have problems if an incorrectly set password policy is in place.
The main advantage of Redis is that it is best in single-tenant environments, such as when a single website uses an entire VPS or cluster of servers. However, you can use Redis on a shared server if you have confidence in the security team of your hosting provider.
Memcached Server with SASL security where each user has its own Memcached instance is a more reliable option for a shared hosting server. Without SASL enabled, one user can erase or alter the data another person has stored in their cache. If Memcached service isn’t using SASL, it doesn’t have security and puts your eCommerce websites at risk.
MechanicWeb offers a security-hardened object cache implementation of Memcached for your eCommerce website on shared hosting. They also provide the fastest servers based on AMD Ryzen Zen 3. We recommend MechanicWeb’ 3 GB RAM/ 3 CPUs Prime plan for eCommerce websites, which is available for $13 per month on Dallas, Los Angeles, or Germany servers. This shared hosting plan will offer great speed for your website, up to ninety per cent faster than other web hosts such as Vultr HF on Cloudways, but will cost you significantly less. You might increase the conversion rate if you made your website faster, so don’t overlook the chance to use this service.
Do not turn on Redis or Memcached without checking with technical support to see if these options are supported. Even if you enable the Redis PHP extensions in the cPanel for connections, what exactly will you connect to?
If your hosting provider does not offer object caching, you should ask them about the OPcache extension in PHP. If they support it (which is almost certain), you may consider using an Object Cache plugin Docket Cache that does not require Memcached or Redis to work. This plugin aims to convert objects into plain PHP code easily stored in the filesystem and loaded from there. This can greatly improve performance and reduce server load. It can be used with other plugins like LiteSpeed Cache as long as the Object Cache is turned off in the LiteSpeed settings.
Before executing a PHP script, the PHP interpreter reads the script from the filesystem, compiles the contents into bytecode, and only then executes the bytecode. OPcache just stores that bytecode in shared memory on the system, and the next request for that specific PHP file (if cached) can skip the reading/parsing/compiling phases, making it significantly faster  at the expense of using some shared memory space.
Since LSCache prepares the cache of all static resources, it is not necessary to run PHP for static pages and therefore there is no need or benefit from using OPcache for simple static webpages. Using OPcache will be beneficial for a busy eCommerce website. By compiling and caching a PHP script, OPcache improves the performance of any non-cached dynamic page.  Combine OPcache with Redis/Memcached for optimal performance when serving dynamic web pages. Small sites can be faster with Memcached. Those that require more resources should use Redis. OPcache speeds up PHP code performance by caching compiled codes in memory, but it requires more RAM, so you run the risk of hitting resource limits. You can enable OPcache and LSCache, but in the rare case if you experience problems with RAM, you should disable OPcache.
Everything runs on WordPress. It has Memcache and the LiteSpeed plugin, and yes, it’s NVMe and OpenLiteSpeed. But even with all that, before OPCache I had a few spikes a day. Now I do not. No spikes in 5 days. My larger sites have moderate traffic, but also lots of editors constantly adding and updating posts throughout the day. They used to make the server crawl. Now: nothing.
OPcache is the hidden gem people often miss.

Don't forget to use a CDN

If you still have a high load on a server, you have to implement a simple mitigation step described in our separate post. Turn on the QUIC.cloud CDN to serve your website. After that, your server will only be burdened by requests to provide infrequently used resources.
Even if this is not enough, don’t rush yet to use the dedicated web server. Choose Ecwid or WP Shopify plugin instead of WooCommerce – and your Shopping Carts pages will be served by their dedicated fast servers, making your website free to deal with the rest of the load.
The majority of readers of this post don’t have enough traffic to worry about OPcache, Redis, and Memcached. They should focus on choosing the right CDN. If you have enough traffic, select QUIC.cloud CDN which is tightly integrated with the LiteSpeed Cache Engine.
Consider using BunnyCDN’s geo-replicated Perma Cache for low traffic sites. Your static assets will be delivered via local BunnyCDN PoPs or directly from BunnyCDN storage centers on each continent, in the case of a cold cache in a local PoP. For such brilliant services, you need to pay only about $1 per month. In addition, you need to set up LiteSpeed Cache plugin to have your static HTML files delivered via free QUIC.cloud CDN. In this case, your origin server will be used primarily for editing your website files and occasionally for delivering cached HTML files that have expired on the CDN. Even if you choose a hosting provider with relatively low resource allocation, you can rest easy knowing that your speed score is largely determined by the BunnyCDN and QUIC.cloud CDNs, as well as by the optimisation of all assets by the LiteSpeed Cache plugin.

Disk I/O speed

Another issue to check is your Disk I/O speed allowance on the plan. It determines how fast your website and scripts are allowed to perform input/output (I/O) operations per second on your hosting server’s disk when a visitor hits your website. If you are throttled to a low disk I/O speed, your website and scripts will always perform slower. It will not matter how much storage, CPU or RAM you have. Avoid values below 30 MB/s. Even 20 MB/s is too low, and you should push a hosting provider to increase it before signing for their plan – or walk away. The current gold standard in our study is 50 MB/s. About 100 MB/s is the most appropriate value for NVMe SSD storage speed, but hosting providers are too conservative, sometimes incompetent, or just unemphatic to pass the increased speed to their users.
CloudLinux recommends using a speed restriction of 4 MB/s for the I/O speed for the HDD drives. It is shocking to find that another darling of so many WordPress bloggers, A2Hosting, is applying such ancient limits of 4Mb/s on their NVMe servers, two generations apart from HDD technology and 50 times faster. The company is followed by Hostinger, who is offering a terrifying 1 MB/s speed on their Business WordPress plan. Hostinger updated its disk speed to 10 MB/s but only on its most expensive WordPress Pro plan (also traded as Woo Premium and Cloud Business Startup) in August 2021.

The end of the customised software era

In your hosting provider selection, don’t force yourself to be restricted to using massive hosting corporations aiming to maximise servers for money. Enthusiasts have created many small hosting companies in the UK and USA with the whole idea to be affordable and fast. They aim to monitor the usage and utilise servers below 60% for the peak traffic to provide breathing space for the growing websites. Their technical support team does not treat clients as just records in the Excel spreadsheet. So, if you behave professionally and politely, they will do everything to keep you as a happy customer.
Please be aware of a strategic problem facing the hosting industry. For the last 16 years, it was dominated by the Nginx (pronounced “engine X”), a specialised web server software developed in 2002 by the Russian software developer Igor Sysoev. It is used by about 50,000,000 websites and is very hard to install and configure on the webserver. All important names in the hosting industry became software developers powerhouses tasked to customise and tune the Nginx. They employed many people and were involved in the acquisition spree to develop and customise, further develop, and further customise. Now all these efforts are waisted – LiteSpeed Web Server has vastly outperformed Nginx. About 5,000,000 recent websites use LiteSpeed now, with the tipping point reached somewhere in 2020.
LiteSpeed Web Server does not require much customisation from the hosting provider, with all development efforts undertaken by the privately-owned LiteSpeed Technologies https://www.litespeedtech.com/. The software uses the same configuration format as Apache HTTP Server and is compatible with most Apache features. LiteSpeed is already an industry-standard supported by most small hosting providers. Larger corporations have a hard time to accept the defeat. They still try to put a lot of marketing effort and directly bribe numerous WordPress bloggers to promote their outdated bespoke tuned Nginx hosting services.
WordPress HTTP/2 Performance by Server
LiteSpeed Web Server boost to the PHP requests processing
In the last decade, it was indeed tricky to select the best server hosting provider. They all ran the home-tuned customised caching and web server software, and some were indeed better than others. Nowadays, everything is much simpler. If the hosting provider offers standard third-party software on their server, it will be your best choice. We have listed all the essential features in the section above, but let’s summarise them again: CloudLinux CageFS, cPanel, Redis, daily off-site backup using JetBackup, Softaculous, Imunify360 or CXS/ImunifyAV+, ModSecurity, MailChannels, LiteSpeed, MariaDB, Let’s Encrypt SSL Certificate, HTTP/3. This list became an industry standard and significantly simplified setting up and maintaining the hosting server. Unless you are in a charity business, you should run away from any hosting provider who cannot accept a defeat and is still mourning about a decade of development invested in optimising the Nginx server.
The hardware of a server is the main difference between modern hosting providers embracing new software suites. Aim for high speed in CPU, RAM and disk I/O speeds to have web servers that are as speedy as possible. Professional third party software and modern hardware – all you need for your website to run smoothly with nothing slowing it down. It has nothing to do with the name or size of your hosting provider. It’s more about what Intel Xeon, AMD EPYC, AMD Ryzen, NVMe SSDs, or DDR4-3200 RAM hardware they offer to run your processes. Be sure that these days any hosting provider is renting a space for their servers at a reputable Data Center fully integrated into global Internet Exchange Points (IXPs).
Nginx vs LiteSpeed
However, as a user, you have to be aware of a reduced barrier to setting up a hosting provider business, encouraging less experienced providers to enter the market. The best modern server can be easily rented from a reputable Data Center. For instance, the best offers from the USA companies listed in our separate study can be replicated in the UK by using the OVH Data Center in London, with a total capacity of 40,000 servers, opened in 2017. They offer servers such as Intel Xeon E-2288G , and Ryzen 7 5800X Zen 3 from AMD. The same dedicated bare-metal servers are also available in the OVH Data Center in the USA, Hillsboro. The fastest server hardware based on Zen 3 AMD Ryzen 9 5950x is available from Hetzner in Germany.
Any person capable of installing third-party software on a server will be qualified to set up a web hosting business. He will provide a good uptime as the server health will be monitored by the Data Center.  The server will have a professional-grade 1 GBps uplink to the network, and you will measure a very fast time-to-the-first byte (TTFB). Everything will work reasonably well as is expected from the commercial-grade third-party software. The quality of the technical support team is the most difficult issue to evaluate. Try to find reviews of the company. If all reviews are positive, it can be a bad sign. You want to have some real-life reviews rather than the one engineered by the hosting provider himself.
The low threshold of setting up the web hosting business implies a low threshold to winding it down. So, hosting with a new company might be riskier than staying with the established micro-business with about five years of serving clients under their belt.

Chapter 2. Cloudways - VPS on shared hardware

Recycling business

Most Data Centers, like Google, are struggling to recycle a significant number of outdated servers. Serious marketing budgets are spent convincing everyone that cloud services are of great value. Ads have flooded Google, Instagram, and Facebook. It doesn’t help that the admins of most Facebook groups related to WordPress communities earn money through affiliate links to Cloudways.  And it doesn’t matter if you get fewer CPU resources on cloud plans than you would from a much cheaper shared hosting provider. The privilege of being served by a household name like DigitalOcean or Vultr outweighs any technical considerations for the uneducated customer.
Cloudways - eye opener
The Cloud Compute Engine service from Google runs on the same infrastructure that Google uses internally for its end-user products. Until recently, it was not the case. External services were recycling old server hardware on the N1 plans. According to Google documentation, customers on their N1 plan use a pool of outdated servers with DDR3 RAM, sometimes eight years old. At the end of 2020, Google has moved most corporate users to N2 or C2 plan machines, which have modern Cascade Lake processors from late 2019, and 2020.
It is ironic to see that Kinsta, who done extensive marketing in 2019 to advertise their ultra-fast Google Cloud N1 servers, is now reporting about 2.5 times improvement in the PHP processing speed on the modern C2 hardware.
PHP processing speed
Kinsta is now reporting about 2.5 times improvement in the PHP processing speed on the modern C2 hardware if compared to N1 servers.
Use this benchmark to think twice about what you read about yet another major darling of the numerous WordPress bloggers, Cloudways. Some of the plans on Cloudways might use outdated hardware similar to those used on N1 servers at Google. According to the sustained CPU endurance test, Vultr is about two times slower than newer C2 servers of the Google Compute Engine. Amazon EC2, DigitalOcean, Linode, and OVH are nearly four times slower. Since Q1 2021, some new accounts on Linode are hosted on already outdated Zen 2 AMD EPYC 7542 servers with 3.2 GHz CPU Turbo frequency (all cores) to avoid such embarrassment in benchmark tests. DigitalOcean also starts offering a premium option for AMD and Intel servers with NVMe SSD. The results of DO Premium are slightly lower than the benchmark results of Vultr HF. However, both are very close to the performance of SiteGround and Kinsta, which use Google Cloud Compute Engine.

CPU vs vCPU

Suppose the website is hosted on a dedicated shared server, under CloudLinux CageFS. Your hosting account has some pre-configured resource allocation in the form of a number of logical cores which can be utilised for your processes. Your processes have access to all threads of the logical core dynamically allocated (as-needed) by Linux. You usually have 100% of this logical core resource. Your hosting server provider assures this via monitoring that overall CPU utilisation on the server is below 60%. Suppose you have heavy web traffic, receiving multiple visors’ clicks requesting dynamic HTML pages, and need to run more than one entry process simultaneously on the same logical core. In that case, it will be done via hardware hyperthreading, and this particular logical core will become utilised to 100%. Multiple threads will slow down each other, but computations will progress for each entry processes via efficient hyperthreading.
Now, lets us understand how a similar scenario is implemented on Cloud services. Cloud services often give access to the single vCPU, which represents a single thread of the physical core. These are virtual CPU cores (vCPUs), each of which is single-threaded and mapped to a hardware processor dynamically (as-needed). Any hyperthreading of the physical CPU is not exposed to the virtual machine. How many threads are running concurrently on the logical core depends on the host node utilisation at any given time. When neighbouring instances consume maximum resources for extended periods of time, it may interfere with your instance’s performance, but the hosting provider does limit such abusive instances to ensure performance for all customers on the host node.
Implications are not good:
  1. If you work on the Cloud server, the overall CPU utilisation is kept close to 100% by selling cheap vCPUs resources for non-time-critical execution, which can be terminated and restarted, balancing the overall CPU load. Your single dedicated thread (vCPU) will be executed on a relatively busy logical core with a good chance of running concurrently with another thread. So occasionally, you will be lucky to have 100% of the logical core for your processes. But frequently, you will get only a 50% share of the logical core resources.
  2. Suppose that on top of all that, you also need to start running several entry processes in your Virtual Machine (VM) instance simultaneously. In that case, they will be co-sharing the same vCPU single-thread resource for all processing. The multiplexing of the calculation resources for simultaneous concurrent entry processes will be performed now via the less efficient Linux kernel software implementation. Your virtual machine cannot utilise a much more efficient hyperthreading technology (up to 16 threads per logical core) and is working in a deprived hardware configuration.
As a result, you should expect a substantially reduced performance in the sustained CPU endurance test on the same hardware configured on the Cloud.
Article appriasal

Shared vCPU plans are not fit for production

Cloudways user are unaware that Linode, DO, and Vultr originally offered shared vCPU plans for developers who needed to spin up a VPS instance for testing, compiling, and prototyping. They were never meant for production use. The majority of blog posts are intended to engage readers and generate referral fees. To obtain a better fee, numerous WordPress bloggers have misled their unaware readers by pretending that there is no fundamental difference between cloud-based shared vCPU accounts on Cloudways and dedicated VPSs. This is very far from reality. There’s a big distinction: with dedicated VPS, your account runs off of dedicated fixed resources such as RAM, CPU cores, hard disk space; whereas following your sign up for an account at Cloudways, they give you access to all these things through virtualisation. Many people think it doesn’t matter or even that cloud services are more reliable – but this couldn’t be farther from the truth!
Cloudways offers shared vCPU instance plans from major cloud hosting companies. The plans allocation is built via a number of provided vCPUs – a single hyper-thread of the physical CPU core – running concurrently with other processes on the physical CPU. The cheap shared vCPU plans are designed by the providers for light web traffic and are explicitly advertised as such on the provider websites.  The shared plans are not guaranteed to perform well. Most of the time, and on recently purchased and still undersold servers, you will have your single thread as the sole process on a physical CPU. However, you must understand the risks involved in using such plans before making any final decision. Your website visitors’ requests are latency-sensitive processes. You have a wide range of fluctuations in the internet traffic and might get some changes synchronously with other accounts on your cloud node – Black Friday is a good example of when sudden rushes can happen for most eCommerce websites.
Both

The epic fail on a Big Day

The cloud hosting provider will never tell you the server’s specification as your account can be freely moved between different servers in the Data Center during maintenance or load distribution activities. This means you’ll never know how much time your website is served using what hardware – some fluctuations may come from chaotic moves between modern and outdated server hardware. Perfomance on Vultr HF varies by as much as 25% depending on specific hardware used for your plan. Good luck with your proactive planning for managing your CPU utilisation before that Big Day!
You risk crashing your website if you don’t have enough resources to execute all HTML requests properly. It is inevitable that when a Big Day comes, you are not ready for the traffic surge. You should be running your website at 25% of utilisation to be prepared, but this may be not enough. When traffic levels reach new heights on Black Fridays, you may notice your vCPU starts receiving less than 50% or even 33% of physical CPU resource availability.  This can be seen as unavoidable, given that every account becomes overloaded simultaneously during this time period. As a result, the physical CPU will serve different accounts concurrently using hyperthreading between their vCPU processes. That’s why many online shops have to scale up their resource allocations by order of magnitude on Black Friday just to cover increased traffic and degraded cloud resources. Your hosting provider will try its best to inject more hardware, but you’re really only a second-grade user, so don’t expect a miracle.
To make things even bleaker, if you reach close to 100% utilisation, your website on Cloudways might temporarily collapse. Parallel processes are running on your vCPU using inefficient software Linux kernel multiplexing. The more processes you have running in parallel on the vCPU, the more inefficient your virtual machine gets. This means that if your server overloads in a short amount of time, there is potential for a snowball effect which could cause your website to go offline temporarily.

Dedicated vCPU plans for production websites

If you ready to pay a 100% premium, companies like Linode, DO, VultrHF, and Google Cloud offer Dedicated or Compute-Optimised plans for their customers. The providers clearly recommend such dedicated vCPU plans for production websites or eCommerce websites. These higher-priced packages will allow a more consistent 100% CPU allocation due to lower host densities and stricter CPU allocations for virtual machines. Anyone who has purchased one of these two options can rest easy knowing they won’t have any problems with sporadic resource stealing by noisy neighbours or overloaded hosts servers. Your hosting provider will be obligated to provide peace of mind for Dedicated plan users and might neglect your struggle on shared vCPU plans. One problem that Cloudways customers have – they don’t get access to the Dedicated plans.
shared vCPU
With the dedicated VPS plans on a dedicated server, you have the same CPU resource regardless of your neighbours’ activities. You will be using efficient hardware hyperthreading on the CPU, which means that overall efficiency does not drop when several processes from your website are served concurrently. So if you had an overload for a short time, that’s no problem at all!
Linode provides the cheapest Dedicated vCPU plans; DigitalOcean and VultrHF are slightly more expensive. Dedicated CPU plans on Vultr are starting from $60 per month. Google Cloud pricing varies depending on usage, with a price similar to Linode at 40% utilisation and significantly more costly than that when used at 100%. All these Dedicated plans are unmanaged – they don’t include any extra features or services that usually come with managed hosting. A fully managed VPS plan from NameHero on the modern server hardware will cost less than even Linodes’ most affordable Dedicated CPU cloud plan. It doesn’t take much math to see how better value can be found elsewhere!
Cloudways vs NameHero

Is it a VPS or a cloud shared hosting?

The DigitalOcean specification is explicit: “…the amount of CPU cycles available for the hypervisor to allocate depends on the workload of the other Droplets sharing that host. If these neighbouring Droplets have a heavy load, one Droplet might get fractional hyperthreads instead of dedicated access to the underlying physical processors. In practice, this means that shared CPU Droplets can have access to full hyperthreads, but it’s not guaranteed.”
Shared instance plans on Cloudways are the cloud version of the shared hosting; but they are priced higher than fully managed VPS hosting on the modern dedicated servers with NameHero or MechanicWeb. With Cloudways account, you share server resources with other customers, and that will cause your website’s performance to suffer when the local cloud nodes get too busy during periods of high traffic load. Dedicated vCPU plans offer something closer to the dedicated VPS plan on dedicated servers with a fixed memory and CPU resource limitation, so your website stays consistent even during high traffic loads.
Good or bad hosting is measured by how well it stands up to a sudden surge of eCommerce traffic that drains your resources with a massive wave of dynamic HTML requests, even if the CPU itself is partially shielded with Redis object cache. For the same money that the customer pays for shared hardware 2 GB RAM / 1 vCPU plan on Cloudways, MechanicWeb provides 6 GB RAM / 4 full CPUs on their shared Boost semi-dedicated plan hosted on a LiteSpeed Web Enterprise modern AMD Ryzen 3 server. Our benchmarks show that MechanicWeb has 1.92 more processing power per CPU core. You don’t have to have a PhD in computer science to tell which setup works better under a peak load.
Due to the flexibility and scalability of cloud hosting, the number of users has increased significantly. However, there’s a catch: you’re not the only one using the hardware – cloud hosting still means shared hosting. Cloud hosting providers give you a vCPU allocation that is shared among all customers of a physical server. This is very similar to how shared hosting works, and if you want to avoid throttling, you need to make sure that your CPU resource usage doesn’t exceed 25% of your limit on average on 1 vCPU/ 1 GB RAM plan. CPU usage should be under 50%, on average, for higher tier plans, like 1 vCPU/2 GB RAM. The rule of thumb is that your vCPU consumption should be about a quarter of your dedicated RAM (in GB). If you have 16GB RAM, for example, your plan comes with 4 dedicated vCPUs. Plans with 4GB RAM get 1 dedicated vCPU. Vultr’s website states that their High FrequencyCompute 4 GB plan includes two vCPUs, so it is still shared. Nevertheless, their 16 GB plan only comes with 4 vCPUs, making it fully dedicated. With their 16 GB plan, you can use any CPU amount within your limits without a fear of being throttled.
Usually, shared hosting providers monitor excessive system resource usage and always contact you to move to the next level if your CPU consumption is too high. Cloud hosting providers are not charities, and they too try to maximise their profits. Their terms and conditions state, “You may not use a shared system in a manner that consumes a disproportionate amount of the system’s resources.” Once your account is found to be using an excessive amount of resources, the cloud hosting provider will begin throttling your CPU allocation. Vultr technical support explains, “…we limit such abusive instances to ensure performance for all customers on the host node.” It is a major drawback of a hosting platform like Cloudways that you cannot communicate directly with your actual hosting provider.
No low-cost VPS provider will sell you a dedicated CPU core for a few dollars a month. I have clients that I manage on VultrHF and DigitalOcean. It’s interesting how often, when you first spin up the VPS, things are running fast and great. Then, a month or two later, the CPU steal is through the roof, and the CPU performance goes down the drain – possibly because the customer is using the CPU so heavily? I often migrate users like that from Cloudways to our shared servers, and customers see an immediate improvement in page load speed.
Cloudways vs LiteSpeed
Now ask yourself: what is a vCPU? I don’t have a problem with virtualization except when used to mask how little resources they’re actually giving you. Sometimes you do get the promised number of vCPU’s, but they’re shared with other accounts.
The partial answer to “What is a vCPU?” question has been provided in the recent report commissioned by Linode. The results show perfect consistency of GeekBench scores on both Google Cloud and newest undersold Linode hosting hardware, suggesting that vCPUs were running independently with other accounts’ workloads. The worst numbers were seen in the AWS EC2 plan, which is an indication that vCPUs on these servers were frequently sharing CPU cores with other accounts. The report also reconfirmed how dismal performance was from both Amazon’s and Google’s side when it came to disk  I/O operations.

Integrated caching creates an illusion of the fast service

WPRocket vs Breeze
Cloudways offers the free Breeze caching plugin that is inadequate. For satisfying Google Page Speed Metrics, an annual fee of $49/site is required to install the WP Rocket cache plugin. Three sites are $99/year. To save that money, you can also try using free versions of SwiftPerformance or PhastPress plugins. The performance should be better than using Breeze.
It is less appreciated, however, that Cloudways’ site uses Varnish caching by default.  Varnish serves as a reverse proxy to externally cache HTTP requests. Once the caching of the content of a webpage is complete, subsequent requests from visitors to the same webpage will not hit the server directly.  Your visitor’s request will most likely be served from the Varnish Cache without even being directed to an HTTP server. It’s very similar to what you would get if you used CDN with the Origin Shield.
Cloudways Varnish setup
Cloudways Varnish setup. https://www.cloudways.com/blog/configure-varnish-cache/
As explained by Cloudways: “… with Varnish proxy in place, your website will be able to manage a huge number of concurrent requests from several users as the server will not even be started. This will result in a magical increase in performance for your application.”
On Cloudways servers, Varnish integration is not the only caching layer. By default, they also provide OPcache, Nginx cache and Memcached, which is used to cache data and objects in RAM to speed up dynamic, database-driven websites. Memcached’s default expiration time is 3 hours. Varnish caches typically expire after 4 to 24 hours, which is a typical CDN cache expiration time. The default time in the Nginx cache is one month. In Cloudways, heavy caching strategies make additional WP cache plugin almost obsolete – the WP Rocket plugin prepares assets, while the actual caching is done elsewhere.
The automatic installation of Varnish cache, Nginx cache, OPcache, and Memcached on fresh Cloudways installs confuses inexperienced bloggers. It’s easy to uncover these caching layers with a tool like https://ismypagecached.com/, but most bloggers don’t do that. It doesn’t matter if you remove all the configurable caching layers like Varnish and Nginx, you’ll still be served by Memcached and OPcache on Cloudways. PHP scripts are compiled and cached using OPcache. All repetitive requests to your websites are sped up as PHP code only needs to be compiled on the first request.
The “professional” bloggers’ benchmarks measure the performance of websites hosted on different fresh hosting installations. To make the comparison fair (!), they don’t configure the caching plugin settings on fresh WordPress installations. OPcache alone reduces the processing time for uncached requests by about ten times. As a result, Cloudways has an unfair advantage over all other hosting providers because their plans include full-fledged caching outside of your WP installation. Cloudways’ “professional” affiliates don’t appreciate that their testing methodology doesn’t include origin servers but instead measures the response time of a Varnish cache web application accelerator.
Apparently, Cloudways is not the only provider that automatically sets up caching layers for bending results of tests performed by “professional” bloggers. Similar activated OPcache caching layers are offered by Kinsta, A2Hosting, WPX, and Hostinger. Take a look at such “professionally” performed benchmarks, and you won’t be surprised to find these hosting providers at the top of the league.
Professional hosting providers rarely use OPcache as their default caching layer. Since no PHP code is executed when serving user requests from the server cache, OPcache has no relevance when serving fully cached static web pages. OPcache might allocate too much RAM to cache precompiled PHP code on a complex eCommerce website. To control the use of RAM, the website owner must be given the freedom to decide whether or not to use OPcache along with Redis or Memcached object caches. Hosting providers who turn on OPCache by default on their servers are aware of the “professional” tests performed by numerous naive and ignorant WordPress bloggers.
If your site uses a lot of blogs and long-form content, there’s a good chance it will attract international traffic. A significant number of visitors will come from outside your country. Google doesn’t know that you are a local business that is only interested in local customers. Websites with international traffic will not be penalised under the new Google Core Web Vitals. However, the data is collected from a global set of Chrome users and the metrics are calculated from that data. Google ignores 25% of the worst data that Chrome reports from your actual visitors, but you still have to provide fast-loading pages to the remaining 75%. Therefore, you should be careful if you are not using a CDN! Page Experience algorithms are not forgiving when remote locations, lack of CDNs, or high server loads slow down the delivery of page content to international visitors.
The QUIC.cloud CDN and LiteSpeed Cache are included as a part of the LiteSpeed Web Server and must be manually activated. Imagine that QUIC.cloud CDN and LiteSpeed Cache are enabled by default. Bloggers who use a “professional” benchmark methodology will achieve excellent results on any LiteSpeed servers. Rather than server hardware, they will test out QUIC.cloud’s CDN capabilities, which are perfect! However, since they have affiliate fees from Cloudways instead of LiteSpeed, they will immediately identify the flaw in the methodology and will turn off all the caching layers on LiteSpeed to provide a “fair” comparison.
Indeed, Tom Dupuis from On-line Media Masters is happy to explain that Matthew Woodward gave a clear advantage in his tests to WPX Hosting because of WPX’s built-in caching (W3 Total Cache) and WPX Cloud CDN. The very same blog post also shows Tom Dupuis’s data that can be seen as providing a clear advantage to the Cloudways due to the built-in Varnish proxy cache web application accelerator, Nginx cache, OPcache, and Memcached. Upon learning about Matthew’s biased data, you might think Tom would be interested in learning what is being tested and compared in his own tests, but he does not care about such details. A typical sign of the amateur affiliate community around WordPress is the lack of interest in learning what a study tests and what a hypothesis is. Studies are primarily done for affiliate fees, and Cloudways offers good rates!
Onlinemediamasters.com

In many cases, LiteSpeed hosting is better than Cloudways

According to our study, LiteSpeed servers are up to 90% faster than Cloudways on Vultr HF. LiteSpeed hosting providers offer from 3 to 10 times better value for money, depending on plan choice. Comparing Cloudways plans with the options shown in our benchmark will help you make an informed decision.
It would be best if you did not rely on Cloudways as a magic solution that offers more resources when your shared hosting plan has reached its maximum resource allocation. The opposite is true! For example, A2Hosting offers 4 GB RAM and 4 CPUs in its Turbo Max plan. Its single-core performance is at least as good as Cloudways. If you switch to Cloudways, your ability to serve visitors might be reduced by up to four times. So, you should expect to handle the peak load of 4x more visitors on the A2Hosting plan. We don’t recommend A2Hosting due to its abysmal disk I/O speed of 4 MB/s, but we are sure that switching to a shared vCPU plan on Cloudways will not solve your problem!

Cloudways - is it a good value for money?

Cloudways falsely advertise the Shared instance vCPUs as full CPU cores on DigitalOcean, Linode, and Vultr. Many users are shocked to recognise that they are served on the cloud shared hosting, sliced, wrapped, and sold under the VPS title.
For a company to claim to be the best while offering five to ten times less value than the competition is not uncommon. But does anyone know of another Western company that pays affiliate fees to people that spread misinformation in numerous Facebook groups? The company that pays affiliate fees to Facebook group admins, which motivates them to remove critical comments about the company or even ban the company’s constant critics from their Facebook groups?
We could not think of another example of such a post-truth world being pushed by Facebook, where fake statements are widely quoted to persuade users to purchase overpriced products and use them outside of the technical specifications of the vendors. Cloudways is crossing ethical and legal lines by using Facebook groups, part of the Facebook ecosystem, to manipulate customers. When we complained, Cloudways just shrugged, “We can’t stop affiliates from promoting Cloudways because they make money doing it, and we’re just three community managers…”
Dedicated-CPU-on Cloudways
Within two hours, Adam Preiser, the group administrator and Cloudways affiliate, removed all traces of this chat in his Facebook group “WordPress for Non-Techies by WPCrafter”. That is not a great surprise considering Adam Preiser (WPCrafter) tainted reputation in the WP community as an affiliate-marketer focused only on his bottom-line.
The main Dragon Island for Cloudways affiliate scammers is the Facebook group “WordPress For Non-Techies by WPCrafter” with 46,000 members. The group is a valuable resource to discuss and learn many aspects related to developing your own website. On the other hand, rare discussions about hosting providers is commented on by several dozens of eager Cloudways scammers. Every now and then, one of them will post “Question? Cloudways or Siteground?” eliciting a chorus howling : “You’ll love Cloudways,” “Cloudways. More control, better value”, “Cloudways anytime”, “Cloudways definitely”, “Cloudways… I use it for my website”, “Cloudways. It’s more flexible,” along with another dozen comments that simply say “Cloudways.”
Adam Preiser, the administrator of the group and the gargantuan Cloudways affiliate scammer himself, fails to prevent the scavenging of the non-techies in his group. He regularly removes negative comments about Cloudways as they are not helpful to his affiliate income. Adam Preiser cowardly deleted users’ comments without notifying them, violating the last civil norms to protect free expression on Facebook.
We have contacted the 100 most active Cloudways affiliates from the “WordPress For Non-Techies by WPCrafter” group and asked them to comment on our research article. No one was brave enough to do it!
This kind of affiliate network is unbeatable!
This kind of affiliate network is unbeatable!
More and more members of another major Facebook group, “WordPress Hosting,” have become fed up with Cloudways scammers in recent months. Nonetheless, this group with 18,000 members has played a negative role in providing Cloudways scammers with a free platform to deceive its visitors. Andrew Killen, the group administrator, treats Cloudways scammers like decent contributors: “Honestly many people just like Cloudways, they seem to have funboys like Apple”. A second group admin, Rob Marlbrough, has also been an active contributor to Cloudways propaganda “1GB Cloudways plan could handle 20 low traffic sites easily. Disk space is the real concern if you want to add even more”.
WordPress Hosting Facebook group
WordPress Hosting Facebook group
WordPress Hosting Facebook group admins are aggressive about silencing or banning members who hold opinions different from their own. The group actively impose the use of shared vCPU plans by calling them VPS accounts. Web designers aren’t informed that the vast majority of affordable VPS plans come with shared CPUs, so they shouldn’t use them unless their developer needs root access. In addition to Cloudways, expensive management panels like RunCloud and GridPane (with added malware scanning) are heavily promoted for shared vCPU plans. GridPane with shared vCPUs from Vultr HF or DO is considered by the group members as the hallmark of a professional web design agency. It is true that many professional designers using GridPane are also experienced WordPress developers using Oxygen Builder. Some of them require root access and therefore benefit from using GridPane.

Facebook WordPress Hosting is the Cloudways group. You’re only supposed to be asking about Cloudways, so people can tell you that you should be using Cloudways!

We observe that the WordPress Hosting Facebook group has created a cargo cult for naive and incompetent web designers. The followers of this cargo cult believe that moving to Cloudways and later upgrading to GridPane will turn them into professionals. We fear that GridPane’s move will not enhance web designers’ knowledge of improving web page load speed or achieving better Core Web Vitals scores. It’s fascinating to learn that the most active proponents of GridPane and shared vCPU plans in WordPress Hosting Facebook groups stick to ThemeForest themes and zealously despise LiteSpeed and Elementor.
Those who think that using DO with GridPane offers something entirely different than shared hosting are making a logical mistake. No, it does not. The basic Droplet offers shared vCPUs. Read the DO’s website – basic Droplets are only recommended for staging, testing, and low traffic projects. With Droplets and GridPane, you can create custom SHARED hosting that costs ten times as much. If you do not need root access, why would you pay ten times as much for it?
When I was looking for a hosting plan back in my pre-web designer days, I didn’t have a clue what I needed. Ubiquitous marketing plays a big part in the decision process, and it sells. After trying a few of the usual (well-marketed) “cheap” providers, I ended up sticking with a VPS service for £39 per month for a couple of years. At the time, I knew it was overkill for my needs (a big understatement), but I honestly didn’t know where to go. This experience has helped me learn more about what I need from a hosting provider:
– Helpful support, – Decent performance, – Reasonable price, – cPanel access, – At least 2 GB RAM, – LiteSpeed server to use the LiteSpeed cache plugin, – Good bandwidth, – Great reviews!

Cloudways intentionally deceives its customers

If you read the following two screenshots, you will appreciate that Cloudways intentionally misleads its users about the nature of shared hosting resources. When you subscribe to their services, you may find a message on the Server Management > Vertical Scaling tab stating that the 1 GB and 2 GB plans are for testing purposes only and are suitable for staging websites. Production sites must be served by 4 GB (or higher) plans. The Cloudways support team advises users to scale up to 4GB RAM Vultr HF if using Redis Cache in WordPress. This is a not-insignificant statement in the fine print that is completely missing from the company’s main website! Many Cloudways users did not plan to pay $50 a month for their hosting!
Cloudways_scaling
Recommended: 4GB (or higher) for production website(s).
2 or 1GB is for staging website(s).
In our opinion, Cloudways offers the worst value for money in the current market. Bloggers are often bribed with affiliate fees to write enthusiastic reviews about Cloudways. Members of Facebook groups are actively instigating users with a large number of websites to move to Cloudways, despite knowing that such an offer is outside the technical specification of the vendors. You can find Cloudways own recommendations to host from ten to twenty websites on a 1GB / 1vCPU plan, to the dismay of more professional members of the Facebook groups. With about 0.5 GB of the RAM used by the Linux, you are left with only 0.5 GB in the RAM to serve your websites. Having 10 websites to share such limited RAM resources is a far cry from anything even the most dubious hosting companies would do!
Moez here from Cloudways reply

Compromised security for Cloudways accounts

The industry standard for cPanel accounts includes malware scanning and web application firewall (WAF) – imunify360/ ModSecurity. At Cloudways, they don’t offer anything like that because it requires paid licences. Instead, Cloudways advise you to use Sucuri’s subsidised subscription for $10 per month plus $16.5 per month per domain for any further domains. It’s an extra expense, but one you are sure to enjoy, especially after upgrading to a $50 4GB plan and paying another $20 per month for WP Rocket as well! Instead of paying $10 per month for much better and faster hosting from MechanicWeb with free LiteSpeed Cache and QUIC.cloud CDN, why not spend more than $100 per month for the slower Cloudways, the inferior WP Rocket, and overpriced Sucuri?
Each deployed server has its own firewall and all your applications are also behind the same firewall with very tight controls. Additionally, we regularly patch the servers that host your applications and we minimize the number of services open to the Internet (minimal attack surface).
For the continuous protection of your application(s), we have partnered with Sucuri to offer you their services (that includes malware cleaning and monitoring) at a discounted price.
Affiliate scammers even suggest you have multiple small Cloudways accounts so you do not put all your eggs in one basket – they know well that it’s easy to hack unprotected Cloudways accounts. If you are dying to get that adrenaline rush and scramble to find out if the latest ransomware attack has compromised your customers’ personal data, Cloudways is your best choice!
There’s a triple trap you fall into with Cloudways. Your WordPress installation is powered by outdated NGINX software, overpriced hosting resources, and compromised security.
Choosing a hosting provider is one of the most critical decisions. Do not forget to look at the big picture: – What kind of support do they offer? – How often do they back up your website? – Can you perform manual backups and restores? The most overlooked issue is – Security and Malware. Keeping WordPress safe from hackers is a challenge. Compared to other web platforms, it requires careful maintenance. What basic level of security does your potential host offer, and how does it help you protect your files and information? Does it offer a service to help you clean your files and databases from infections? Is there a charge for this service? How long does it take to get rid of the infection after the incident is reported? It is advisable to find out about these timeframes and costs before you decide to host your website(s) with this provider!
Mark Borst
Mark Borst

Conclusion

We’ve done our best to show our readers the true nature of Cloudways and its affiliate community of Facebook scammers. We offer detailed arguments of why we think it’s a good idea to stay away from the company. We would hesitate to compare Cloudways to even the worst EIG hosts. Cloudways misinforms its users about the shared hardware resources offered in its plans. This looks dubious under European consumer law. Choosing the wrong hosting option, which may be unsuitable for your eCommerce website, can affect the user’s experience. As a result, not only will you lose some users due to the drop in conversion rate, but you also run the risk of losing your customers’ financial data due to the threat of an imminent hacking attack on your unprotected website.
The Wix has brilliantly parodied WordPress in the timeless classic trashing video campaign. The story starts with the hilarious video and continues to YouTube. In his blog, Wix CEO wrote: “I mostly care about WordPress’ security. WordPress is so popular; I’m sure my credit card details are on many WordPress websites. It’ll be great if WordPress fixes security so we can all feel safer.”

If only he knew about the irresponsible and reckless affiliate scammers surrounding Cloudways!
At least two community managers from Cloudways, Saad Khan and Aleksandar Savkovic, have read our research article. We have offered to publish Cloudways’ written response to our arguments,  and we have received their verbal commitment to do so in May 2021. This text implies that Cloudways has not yet issued an official statement. Most probably because they have nothing to say in their defence!
Rapid change from "SiteGround" to "Cloudways" fan group
Rapid change from "SiteGround" to "Cloudways" fan club. WordPress Hosting Facebook group.

Chapter 3. How to conduct a benchmark study

WPPerformanceTester plugin

The original publication from Matthew Woodward states, “All previous tests were served from cached results, but the WordPress Performance Tester plugin was designed to test the website performance of the actual server itself for each WordPress site.”
WordPress based benchmark tools is exactly the missing link in testing the server! Currently, we have limited number of free tools to measure the performance of the server. WP Perfomance Tester plugin, however, is one that you should avoid! Kevin Ohashi developed this plugin by combining open-source scripts that were publicly available. Unfortunately, he replaced one of the functions in the script and now his plugin gives huge 75% advantage to AMD CPUs.
Hoping to spread objective benchmarks instead of unsupported marketing claims, he also put his WP Performance Tester plugin in the public domain. In addition, Kevin Ohashi invites WordPress hosting companies to pay him for the privilege of participating in his annual testing. Taking part in such annual surveys, using the WP Performance Tester plugin among other tools, is a regular occurrence for Cloudways. Therefore, Cloudways supports the use of this plugin. It is amusing to see that AMD Ryzen Zen 3 shows benchmark performance 3.9 times (!) faster than Vultr HF with WP Perfromnce Tester, even though Cloudways boasts of providing the fastest service to its customers.

Problems with WP Performance Benchmark

We installed the WP Performance Tester plugin on dozens of dummy websites to measure PHP benchmarks on different servers. The results of our study were compared with those of Kevin Ohashi from 2021 and data from Matthew Woodward. The plugin performs two types of tests. The “Server Performance Benchmark” tests the hardware capabilities of the server to run PHP code. The second, called “WP Performance Benchmark”, tests the WordPress interface’s latency when interacting with a database via the $wpdb class, bypassing SQL.
Based on our data and data presented by Kevin Ohashi, we found no correlation between hardware performance tests and latency tests. Hardware benchmarks correlate well with changes in CPU speed between different servers. However, the latency of interacting with a database differs significantly between different servers. It is not affected by caching, CPU speed, CPU cores, amount of RAM, or disk I/O speed.
WordPress Performance bencmark measures the latency of requests to the database. It is not related to the hardware performance of the server.
WordPress Performance benchmark measures the latency of requests to the database.
It is not related to the hardware performance of the server.
The data from Kevin Ohashi show no correlation between the two components of the test, while the data from Matthew Woodward show a near-perfect correlation. Consequently, we have found that Matthew has completely falsified his data. It’s not the first time people have published fake data, but it’s revealing to discover the obvious fraud in the data published by the well-known SEO blogger.
We expressed concerns about the WP Performance Benchmark part of the plugin tests in our conversation with Kevin Ohashi. We advised him to change the test title. It is puzzling that the WP Performance Benchmarks has changed twofold on Cloudways plans on two separate instances with DO Premium 2GB.  However, the Server Performance Benchmarks were identical. The problem is not isolated. The WP Performance Benchmark varied by a factor of 2.5 across three types of servers provided by MechanicWeb from Los Angeles, Dallas, and Amsterdam, implying a latency variation of 0.6 ms to 2 ms.
However, MechanicWeb discovered the cause of this latency variation and reduced it to almost zero. In one of their data centres, WP Performance Benchmark had changed from 440 to 2801, more than a six-fold increase! And now it’s two times better than the best score from Cloudways, giving MechanicWeb the top spot among all other hosting providers participating in Kevin Ohashi’s annual surveys. Of particular note are the excellent results that MechanicWeb plans achieved in the Server Performance benchmarks. Their results were four times better than those of Vultr HF on Cloudways and two times better than the best results ever recorded by Kevin Ohashi.
Were there any improvements recorded by MechanicWeb after fine tuning? The largest change was for TTFB, which was reduced to 68ms. The overall loading time for a dummy website dropped by 150ms. The fine-tuning unleashed the hardware performance by removing delays in processing requests to the WordPress database. MechanicWeb was pleased with the test plugin as it allowed them to identify the problem and gave them a measurement tool to check for the bottlenecks on the server to fix the problem. The optimised server is capable of serving larger website traffic. This was confirmed in load tests using K6.io and Loader.io testing tools. However, WP Performance benchmark results cannot quantify the extent of improvement. It will very much depend on your specific website. We believe that a more holistic test such as the K6 load test for the uncached dummy website described below gives a better measure of the server’s overall performance.
The real-world performance of WordPress is not as severely affected by latency as implied by benchmarks. On servers with heavily throttled disk I/O speeds like A2Hosting with its pathetic 4 MB/s, only a meagre 2.5 kB of data is read or written in the 0.6 ms time span that corresponds to its database request latency. We don’t want to put that much emphasis on the exact values of the short latency of the database requests because it’s only one part of a more complex picture. This score surely provides a great indicator of the sysadmin team experience in fine-tuning WordPress performance settings, but it does not go much further.
The WP Performance Tester plugin provides usable results in the WP Performance Benchmark section to help a hosting provider fine-tune their server. However, it is not as useful to us for comparing different hosting providers. By measuring the latency of interaction with the WP database via the $wpdb class, the WP Performance Benchmark does not provide insight into the average performance of hosting providers in daily use. Using such a test to evaluate the WordPress software stack is not appropriate. Delays can be caused by settings in CageFS, imunify360, ModSecurity, MySQL, etc. Depending on the design of the benchmark test, delays may affect benchmark results differently from how they affect your site’s performance on dynamic queries.
Based on our analysis of the raw data, we have concluded that the second part of the plugin test, the WP Performance Benchmark, is an important but indirect and uncertain contributor in comparing hosting providers. Consequently, we use it mainly to benchmarks the quality of the sysadmin team in fine-tuning MySQL and other aspects of the hosting.  Unfortunately, the first part of this benchmark plugin,  the Server Performance Benchmark, has its own bug.

Problems with WP Performance MySQL test

Upon closer inspection, we found that the so-called MySQL query benchmark test is using a specific internal MySQL function and is heavily biased in favour of the AMD architecture. Since not all real-world processes are affected in the same way, this benchmark has its own limitations when comparing Intel servers to AMD servers. To establish the actual degree of the bias, we have carried out a cross-comparison of results measured using WPPerformanceTester plugin,  WPBenchmark plugin, and original Benchmark.php script. Kevin Ohashi adapted this script for his WPPerformanceTester plugin but changed a specific MySQL function used by the test. The results of the WPPerformanceTester plugin’s MySQL query test are shown below in comparison to the aggregate CPU benchmark, which is the result of averaging three independent benchmarks.
Benchmark_Intel_vs_AMD
While all the Intel-based vendors have fallen into a neat line, the four AMD-based hosting providers have taken a different path. With our knowledge of the exact server specifications used by the vendors, we can surmise that their aggregated CPU benchmark scores are correctly position them together with the Intel-based servers with a similar CPU frequency. This chart illustrates the significant, 75% performance gain for AMD when it comes to specific MySQL function selected by Kevin Ohashi.  In real-world applications, however, these benchmark results should not be trusted without authoritative investigations. For our combined benchmark tables, we multiply the WP Performance Tester MySQL test results for AMD-based servers by 1.75 to correct for the artefact.
Please be aware that AMD EPYC performance boost is not universal for all MySQL functions. The WPPerformanceTester plugin uses the ENCODE() function to measure the so-called MySQL query test, which gives an estimated boost of about 1.75x to AMD. Using the AES _ENCRYPT() function of the original MySQL Benchmark.php test removed the boost for AMD servers. The figure below does not show a breakdown by Intel and AMD architecture. This means that the boost for AMD servers is an artefact built into its hosting benchmark plugin by Kevin Ohashi. In other words, please be aware that benchmark plugin developers don’t care about performing thorough testing and analysis of their algorithms.
Kevin Ohashi deserves some criticism, followed by praise. The WPPerformanceTester plugin is an indispensable tool for testing hosting providers. With it, you can quickly tell if you are dealing with AMD or Intel servers. Without that significant artefact generated by Kevin Ohashi, it would be almost impossible!

WPBenchmark and Benchmark.php tests

We have modified original  Benchmark.php  file – an open-source script which was copied and changed for WP Perfomance Tester by Kevin Ohashi. We have replaced the original string with the:
$query = “SELECT BENCHMARK(10000000, AES_ENCRYPT(‘hello’, RAND()));”;
We increased the number of tests tenfold and added the RAND() function. Our goal was to get Benchmark.php to perform a measurement as similar as possible to that of the WPPerformance Tester plugin. The only difference between the two tests now is the functions AES _ENCRYPT() and ENCODE().
Benchmark_PHP
We used the result of the “operations with large text data” test in the WPBenchmark plugin as a third cahnnel for aggregated CPU performance. The Benchmark.php tests appear to accurately measure the performance of CPU, but the WPBenchmark test results are slightly skewed by the software delays inherent in a given server.
WP_Benchmark
We corrected the AMD test results measured with the WPPerformanceTest plugin and calculated the aggregated CPU benchmark value as an equaly weighted average of the three benchmark channels. The aggregated CPU benchmark was normalised to the value measured for the Vultr HF server ordered from Cloudways. The hosting provider CPU scores are expressed as deviations from a Vultr HF reference value in percent. We calculated that the CPU score of the Ryzen Zen 3 server is 92% higher than that of the Vultr HF server. Raidboxes servers have a 28% lower CPU score than Vultr HF. Compared to Raidboxes servers, Ryzen Zen 3 server is (1+0.92)/(1-0.28)=2.67 times faster.

Independent AMD EPYC tests

Our conclusion about unqalified boost for AMD servers was put through its paces with a Google search to see if it stood up to professional comparison. We were pleased to find a recent report, “CPU performance in shared hosting environment”, from the CloudLinux team that examined the same issue. To avoid similar embarrassments shown with our WP Performance Tester plugin analysis, the CloudLinux team created a professional representative mockup of the shared hosting environment with CageFS. They put 200 user accounts on each server, installed WordPress for each user, and created 12 unique pages on each domain.
As a result, the research team obtained 2,400 unique URLs to test server load. Random requests were sent to 2,400 URLs. It took 5 minutes to complete a test iteration. Each iteration had 150 concurrent, non-keep-alive connections. That is approximately two concurrent users per logical CPU core. At the end of each iteration, the total number of requests processed and the number of requests per second were calculated. The spread between different servers was between 12 and 16 requests per second per logical CPU core.
CloudLinux team concluded that the AMD Rome (Zen 2) architecture offers from 0% (for 7302) to 18% (for 7542) better aggregated CPU performance, while AMD Milan (Zen 3) offers 28% (for 7313) more CPU performance.  These numbers are derived by correcting the raw data by the actual CPU speed of 2.9 GHz for Intel Xeon and 7542 vs 3.0 GHz for 7302 and 7313. The researchers also measured a 9% performance boost for LiteSpeed web servers in their benchmarks for web servers and PHP handlers. The numbers above are significantly lower than those claimed by the LiteSpeed team, but this integrated metric provides a more accurate comparison. The researchers found that the new AMD EPYC offers a 32% lower price per request, suggesting that more hosting providers switch from Intel to AMD EPYC in the coming years.
AMD gain of function
Maximum requests per seconds on 32 CPU core servers
with the same frequency on CPU and RAM
as measured by CloudLinux team

K6 load test - a part of our benchmarks

Running free tests provided by K6 is the easiest way to reproduce the CloudLinux trial described above. A dummy website that can be replicated by anyone is created by installing WordPress, Elementor, and Hello Theme. The web pages should not be cached. This forces the server to compile PHP code and provide HTML, CSS and JS code after extracting the data from the database. Anyone can rebuild such a dummy website and run up to 50 tests with a free K6 account. We used 5-minute tests where the number of virtual visitors was ramped up during the first minute and ramped down during the last minute of the test.
Because Cloudways does not allow OPcache to be disabled, their scores for this test were higher than they should have been for a fully comparable test. OPcache typically increases the results of such tests by a factor from five to ten. To reduce OPcache’s efficiency, we reduced its memory limit to the minimum allowed value of 9 MB. Nevertheless, it is reasonable to assume that the Cloudways results are overestimated.
By changing the number of virtual visitors, the average response time per visitor can be increased to over one second. Under such conditions, the number of virtual visitors served per second reaches a saturation point. As the number of concurrent visitors increases, the response time per visitor increases, and as the number of simultaneous visitors decreases, the response time per visitor decreases. However, the system throughput (number of visitors per second) remains the same. Such saturated system performance – the number of visitors per second – is proportional to the CPU allocation. The cPanel resource monitor shows that under saturated conditions, the K6 tests use 100% of CPU capacity.
In the following graph, we compare the performance with the monthly fee paid to the hosting providers. The price-performance ratio can be easily seen from this graph. There is no doubt that LiteSpeed hosting providers are significantly better than Cloudways. Ten times better value for money is no small thing compared to the claims of Cloudways affiliate scammers!
LiteSpeed_vs Cloud
For more details on the LiteSpeed web hosting providers examined in this price-performance analysis, see our separate report.

Results of our Benchmark study

Faster website performance can be worth more money in some cases. What then can be found out about the speed difference between different hosting providers? As discussed in the previous chapter, we use two free WP plugins, such as WP Benchmark and WP Performance Tester, to run tests. Additionally, we used Benchmark.php – an open-source script which was copied and modified for WP Perfomance Tester by Kevin Ohashi. Benchmark.php was used because it uses MySQL functions that do not favor AMD processors. The KeyCDN Performance Test was also used to test TTFB at the geographic location closest to the hosting server.
Top_LiteSpeed_providers
The table lists some of the best LiteSpeed hosting providers. The “MySQL function” test for AMD servers has been corrected by multiplying by 1.75 to remove artificial bias; these data are marked with the asterisk.
Top Nginx providers
We do not recommend these hosting providers.  We just want to show you benchmarks of the best Nginx hosting services. Cloudways’ test results are overstated because it is impossible to turn off OPcache; we reduced the memory allocation to the minimum allowable value of 9 MB for OPcache to reduce the boost. Raidboxes is inefficient and slow at delivering non-cached requests; we question the usefulness of using such a service.
All-incl, Nexcess (Liquid Web), SiteGround, and 20i do not limit CPU allocation, so the K6 test does not show saturation for the number of visitors served. SiteGround limits your account by total monthly visitors, while 20i limits it by the overall monthly bandwidth. WP Engine shows highly dissapointing results. Their caching layer cannot be removed, but WP Engine will start rate limiting your site if the number of visitors exceeds 30 per second.
Second_Tier LiteSpeed_providers
To help you choose the best LiteSpeed hosting provider, we have benchmarked more companies. We do not recommend any of the hosting providers listed in this third table. At maximum load, Closte generates errors. WPX generates errors and experiences sporadic delays of TTFB up to eight seconds at just two visitors per second. Benchmarking studies on Hostinger and GreenGeeks show them to be pure junk-grade providers, yet they are enthusiastically recommended by a large number of illiterate affiliate scammers.

The importance of bypassing caching

Most LiteSpeed vendors provide access to PHP extensions through the cPanel, so caching can be easily configured. However, with many of the companies compared above, caching settings are hidden behind proprietary control panels. We always added the code recommended by A2Hosting to the .htaccess file to turn off hidden caching levels on the Apache webserver. If you don’t have access to the files on the server, you can use LiteSpeed pluging to edit the .htaccess file. Alternatively you can use WP File Manager plugin.
In some cases, such as WPX, this works well. Some caching levels, like OPcache, remained active in other cases, such as KnownHost. Based on the K6 load test result, we can see how the response time increases with the number of virtual visitors served by the K6 loader. If this time does not increase under heavy load, it can be interpreted as a sign that some caching layers are still active. The number of users served per second should be saturated at a reasonable level while the CPU runs 100%. This is a situation where you can be sure that you measure the uncached capabilities of the server.
It is essential to recognise how our tests differ from those of uneducated affiliate scammers. We take extensive steps to disable caching layers and confirm their removal. We test server performance instead of external caching layers like Varnish servers or CDNs. K6 load and TTFB tests can give you an idea of server performance on dynamic requests for non-cached resources. LiteSpeed server performance when serving fully cached resources is comparable to Varnish cache. On LiteSpeed, the TTFB values for cached resources are even lower. Whenever we fail to disable caching on a particular server, we record it. We are not telling you that this server is better than any other on the market. Not at all! In its determination to prevent us from changing caching settings, this hosting provider outperforms all of its competitors. That is the only achievement of this provider!

Our findings

Unsurprisingly, Cloudways is not the top performer in our benchmark tables. It’s just as good as NameHero, Kinsta, Nexcess, A2Hosting, and SiteGround. The performance of companies like Nexus Bytes, EthernetServers, NameHero, Nexcess, Snappy Host, WP Engine, 20i, and Raidboxes is degraded due to unoptimised MySQL settings. As a result, these companies appear to offer the worst performance scores in the “Database” and “WordPress Performance” sections. This can be cause by the settings created during the high availability cloud shared server deployment, so reality might be more complex than just using unprofessional sysadmin. Some of the companies above don’t have in-house sysadmins and rent resources from a Data Center, who configures high availability cloud shared servers, the CloudLinux, and CentOS 8 installation themselves.
This is quite an academic article with a lot of data.
I have always tilted towards Cloudways due to its heavy recommendations and popularity. This article is an eye-opener! I will have to give LiteSpeed and QUIC.cloud CDN a try.
Thanks for the invaluable and detailed information provided here. I am sure I will have to read this again, bookmarked!
Nexus Bytes offers affordable and fast AMD Ryzen 2 servers. Due to limited security, they cannot be recommended for general users. By using DirectAdmin instead of cPanel, Nexus Bytes cut corners. They do not replace Imunify360 with an alternative malware scanner. It is very similar to what you have on Cloudways, All-incl, or NameCheap, who also doesn’t offer malware scanner. Therefore, Nexus Bytes is only recommended for the brave users of Cloudways. Hosting services like this are ideal for users who have long ago decided to ignore such nonsense as malware protection.
Sewer.io and MechanicWeb are the winners in the speed test. Using modern workstation-grade servers helped both companies to the top place. MechanicWeb uses Zen 3 generation servers, which are more than 45% faster than Serwer.io’s Zen+ generation servers. MechanicWeb’s Intel workstation-grade servers are only 10% faster than Serwer.io Zen+ servers. Server performance has been radically improved with AMD Ryzen Zen 3. One of the first hosting providers to benefit from this significant increase in server speed is MechanicWeb.
Kevin Ohashi’s credibility was further undermined by a terrible performance by Raidboxes plan for providing uncached resources. In his latest review, Kevin Ohashi recommended both 20i and Raidboxes as top-notch shared hosting providers. While 20i is not the worst option, it is certainly far from the best. We would definitely put Raidboxes at the bottom of any list.
What else did we learn from our study? Scores on different instances of Vultr HF can vary widely. Vultr HF offers three different types of servers. The first type is a cone – it is only 4% faster than normal Vultr servers and 3% slower than DigitalOcean Premium Intel servers. The second type is 7% faster than DigitalOcean Premium, and the third type is 17% faster than DigitalOcean Premium. We tested six Vultr HF servers and found that three of them were slowest type and two were fastest type. Vultr HF plans may experience erratic movement between machines in the Data Center, which can affect your site’s performance. We compared LiteSpeed server providers with the slowest Vultr HF servers.

But what about Closte?

As part of our benchmarking tables, we featured Closte, a unique LiteSpeed service provider that uses the infrastructure of Google Cloud Compute Engine to offer a Pay-As-You-Go hosting plan. Unlike Cloudways, Closte does not compromise on security and offers highly efficient real-time malware detection protection that is tuned for WordPress.
Closte is definitely worth trying. It offers LiteSpeed, security is included, it runs on Google Cloud, and there are no affiliate programs :). Closte costs me $5 a month for a basic website but outperforms my previous providers. With Closte, I have never had a website crash, I have never experienced a hack, and I have never contacted support about anything WordPress related.  However, Closte is not user friendly for newbies. Closte is perfect only if you already have experience managing a WordPress website and know what you are doing.
I am reaching out again regarding your articles – I have studied them, and thank you for giving me and others the privilege of broadening our horizons. I see now that I put Closte too high on a pedestal because they offered me my first look at LiteSpeed servers and the opportunity not to be undermined by other hosting platforms with their drawbacks, mainly security issues. I am migrating to MechanicWeb with a Quic CDN, and Ecwid provides the e-commerce side for a secure hosting solution. Much more has happened due to studies and implementations on my part, but I felt I owed you some honest feedback related to your brilliant, carefully tested articles.
Upon closer inspection, we found that Closte uses similar hardware resources as the low-cost EthernetServers provider. On EthernetServers, the cheapest Silver plan costs $3 per month, which includes five domains and 15GB of storage on the fast NVMe SSD hard drive. For a very low traffic site, Closte costs at least $5 per month, so EthernetServers is the better deal. Closte is notorious for its abysmal customer support, so it’s unclear if their offer is worth the hassle.
Additionally, we recommend considering the junior Smart plan on MechanicWeb servers as an option. They offer 10 GB of storage and two domains for the same $3 per month, but their CPU processing speed is 67% faster than Closte’s. It seems illogical to switch to Closte when the faster processing of dynamic queries on MechanicWeb servers can increase the conversion rates of your visitors.
Hardware resource price on Cloud Hosting is significantly higher than hardware prices for the bare-metal servers at the same Data Center. A higher cost comes from complete virtualisation services with live migrations of Virtual Machines controlled by the engineers at the Data Center. It allows them to better address patching, repairing, and updating the software and hardware without the need for you to worry about machine reboots.
Thanks for the superb analysis! I have to admit that I have been quite a proponent of Cloudways, with their generous affiliate program as the main reason why. It helps that Cloudways pretty much excel in marketing themselves as a modern managed WordPress hosting provider. Without a closer look, not many WordPress developers, including myself, will realize that Cloudways would not be the actual dedicated VPS hosting. So, kudos to your research and this great article!
I feel like this article is a wake-up call for me to move into the future of faster web servers with LiteSpeed. The combination of NameHero servers in the northeastern US and MechanicWeb in the southwestern US seems like good coverage. You mention providers in Europe and I noticed that EthernetServers offers hosting in Sydney and Singapore. Closte has a very high potential among all those discussed in the article. They provide very efficient geo-coverage, and are useful for all my Asian clients.
Again, marketing is everything. It’s challenging to determine the best when we keep getting re-targeted by the big companies. So as web professionals, we need to be able to point out the benchmarks of the large slow providers our clients are familiar with versus the ones championed in your article.
The case for considering Closte services remains strong if you have customers outside the major European and American hubs. By using Closte, you can physically place your server much closer to your customers, drastically reducing their wait time for dynamic requests or even static content on low traffic websites. Closte has locations in Finland, Sao Paulo in Brazil, and many Asian locations, including Seoul, Tokyo, Mumbai, Taiwan, Hong Kong, Jakarta, Singapore, Sydney, and Melbourne. Closte will undoubtedly be the best choice for many businesses in Asia and South America.
There is no doubt that the Cloud Server is the best concept for rarely visited websites with computing resources allocated only when used. However, it is far better to host your website on a much faster entry-level dedicated LiteSpeed Web Server Enterprise shared server plan than choose more expensive Cloud server plans.
Awesome-testimonial

Hosting company size is its liability

Most large hosting corporations are fallen into the double trap. They are the primary Cloud services providers with the most expensive hardware resources, offering outdated bespoke software for their users.
Why are the customers still staying with them? They probably have a more urgent need to hire a knowledgeable web designer to refresh their WordPress website full of bloated code  – but they ignore both issues because they don’t care enough.  Ironically, Google wants them to start caring more about the ultimate page loading speed. In August 2021, many users will find their Google ranking crashed due to low Page Experience signals for their web pages. It is a painful way to force people to look around and find the ugly truth about large hosting corporations – they just lost track.
Now is the critical time for smaller hosting companies to offer better service to the customers. It is easier for smaller companies to understand their clients better. And decide to throw away all bespoke software development performed to optimise the Nginx servers. As they are a small team, they can make such a decision with less hesitation. They have a smaller number of clients compared to larger companies. The same people provide support to the same clients. In time the support team gains a better understanding of client requirements and show more empathy for resolving their problems.
Don’t be over-enthusiastic to choose a large company. It would help if you appreciated the scale of the problem facing large hosting corporations:
  • RackSpace and GoDaddy employs more than 8,000 people each
  • Newfold Digital (parent of Hostgator and Bluehost) employs more than 3,000 people
  • Endurance International Group (EIG) and OVH employ more than 2,000 people each
  • Ionos employs more than 2,000 people
  • WPEngine, Fastly, and NameCheap employs more than 900 people each
  • SiteGround and Hostinger employs more than 600 people each
  • LiquidWeb employs more than 500 people
  • Iomart, InMotion, and WebFlow employs more than 300 people each
  • DreamHost, Flywheel, Kinsta, Cloudways, HostPapa, Fasthosts, Media Temple, and A2Hosting employs more than 150 people each
  • Namesco, HiVelocity, 10Web, Hetzner, and Pantheon employes under 100 people each
With the current price pressure to stay under $20 per month, LiteSpeed plans in the $10-$15 range with cPanel, NVMe SSD drives, and great customer service are the best choice. In a race to the bottom, development teams will need to downsize. Making a brave change to their outdated business model is painful. It is a big personal drama for many managers and software developers. Its something you never want to experience yourself. It is a coming bloodbath. Not all hosting companies can survive for the next few years.
Choose a smaller lean company offering you a combination of modern hardware hosted in a large Data Center and a fully standardised software package – and you can be sure that you will be among happy customers. Chose a large hosting corporation – and you will face what will happen to their technical support team’s moral once the axe starts swinging inside the headquarters.

Chapter 4. Advanced Hosting Options

Premium shared hosting

Congratulations! Your website just became a huge success, as it managed to attract thousands of visitors per day. The first thing you need to do is unload all your static files onto an inexpensive QUIC.cloud CDN. This way, the traffic for the static assets will no longer be directed towards your origin server. This will cost you about $10 per month for 1TB of CDN traffic, but a Pay-As-You-Go version won’t likely exceed more than a $1 monthly fee. The origin server’s response time will be reduced as it left to respond only to requests by visitors who are viewing dynamic HTML pages in their browsers while scrolling through your online shop, reading comments on blog pages, or visiting the membership-only pages.
Transforming your WooCommerce store into a standalone back-end office hosted by a third-party service provider and an independent WordPress front end is the best way to deal with exponentially increasing traffic. You can do this using the Ecwid or dedicated plugin WP Shopify instead of WooCommerce, as we have written about extensively in the separate post. This approach will take over the task of serving checkout requests for you while reducing the load on your origin server so that it doesn’t get bogged down under increased demand from dynamic HTML pages generated by online shoppers visiting your website. Using either Ecwid or Shopify back-end, you are protecting your site’s sensitive data from any potential future hacking. If hackers were to destroy your website in a cyberattack, then all of your customers’ information would be safe on an independent back-end server maintained by either company.
eCommerce with WP Shopify
WooCommerce vs WordPress with WP Shopify plugin
Upgrading your shared hosting account may be a good idea if your traffic levels remain high. In a separate article, we describe the best hosting companies in the UK and US. In the article, we detail MechanicWeb and its seven hosting plans with progressively increasing resources. Upgrading to the Elite package at MechanicWeb offers the best performance. You get 8 CPU cores and 12 GB of RAM on the fastest AMD Ryzen Zen 3 server. You pay $54 per month and get unlimited domains, 200 GB of NVME SSD storage with 250 MB /s speed. Such a configuration outperforms almost any VPS, and you do not need to maintain any software. By using QUIC.cloud CDN, OPcache and Memcached, you can serve an eCommerce store with up to 200k visitors per day without any problems! MechanicWeb offers servers in Dallas, Los Angeles, and Germany. It is the perfect solution for companies based in the UK, Europe, Israel, India, the US and Latin America.
For EU or UK-based traffic, we recommend considering Serwer.io‘s WordPress Delux plan from servers in London or Germany. It is priced at $28 per month and provides 8 CPU cores, 8 GB RAM, 100 simultaneous entry processes, 10 domains, 40 GB NVMe storage (disk speed 64 MB/s), 3Gbps network uplink, enabled Redis, and support for HTTP/3. This configuration is sufficient for several heavily loaded eCommerce sites and outperforms most entry-level VPS offerings. It features workstation-grade AMD Ryzen Threadripper 2950X processors (turbo speed up to 4.4 GHz) with ECC DDR4-2933 RAM. This option is providing more recources than MechanicWeb plans for the same monthly fee, but is based on 45% slower AMD Zen+ hardware.
For much less demanding website you can consider NameHero, the only large provider of LiteSpeed hosting services on our list. If you are looking to host your server in the US, we highly recommend their Business Cloud plan. It provides 4 GB of RAM, 4 CPU cores, 100 simultaneous entry processes, unlimited NVMe storage (disk speed 50 MB/s), 1 Gbps network uplink, enabled Redis, support for HTTP/3 protocol, and MailChannels. Their server is based on AMD EPYC 7302 with 3.3 GHz CPU turbo speed and DDR4-3200 memory. Which is an excellent deal considering it costs only $13.5 per month if you pay for three years upfront! However, MechanicWeb offers nearly twice the speed of NameHero. A fast site can improve conversions, so the tradeoffs do not always work out in NameHero’s favor!
NameHero appriasal

Shared hosting is now offering more resources than before

A hosting provider will monitor that his servers are underutilised to avoid slowing down your neighbours hosted on the same server. Different providers use various metrics for establishing an Unacceptable Resource Usage policy. For instance, at NameHero, the red flag will be shown to your account if any allocation limit is reached for longer than five minutes. In that case, the CloudLinux LVE will generate a snapshot of the running processes and other information to identify what’s using up resources. It helps the support team to troubleshoot resource issues and discuss them with you. NameHero further states: “We usually recommend moving to a VPS from a shared hosting account if the resource consumption on your account is negatively affecting your customers and you’re looking for faster speeds overall.”
MechanicWeb states that a user should not use 80% or more of the system resources for more than 60 seconds. It looks like you may not have dedicated resources, but you’re not severely limited. The requirements stated by NameHero and MechanicWeb are a significant relaxation of the commonly used rules for shared hosting accounts: “An account should not use more than 25% of its CPU allocation continuously for more than 5 minutes”.
When you compare NameHero and MechanicWeb’s generous shared hosting plans to traditional shared VPS providers, you might be confused. The typical resource usage policy for affordable shared VPS plans is 30%. The average usage of some cheap VPS providers is as low as 10%.
Shared VPS providers can force you to upgrade if you use more than 30% of the resources listed in the package specification. NameHero and MechanicWeb, however, do not limit your activity that severely. As a result, their users would consume twice as much CPU power as traditional shared VPS users with the same number of CPU cores. Due to the use of a modern Ryzen Zen 3 processor, MechanicWeb will be able to double this number to four times as much CPU processing power.We hope that such a benchmark will protect you from the influence of followers of the VPS cargo cult.

Reseller hosting plans

Reseller Hosting is available from many providers.  Both MechanicWeb and NameHero are serving web agencies wishing to offer branded hosting to their clients. The services are similarly priced and offer a variety of hosting options, ranging from 20 to 250 accounts. To make the best decision for your clients, be sure to review the details carefully.
Let’s look at the reseller plans billed at $42 per month and compare them to a similarly priced Cloudways plan with 4 GB RAM, 2 CPUs, 80 GB of storage, and disk I/O speed of 120 MB/s on VultrHF servers.
  • NameHero offers 150 GB of storage, 80 cPanel accounts, and free billing automation with a WHMCS or Blesta license. Each child account has an allowance of 2 CPUs, 2 GB RAM, and disk I/O speed of 32 MB/s.
  • MechanicWeb offers 100 GB of storage, 75 cPanel accounts, disk I/O speed of 30 MB/s, 2 GB of RAM, and 1 CPU, which is two times faster than NameHero’s. An additional $10.90 is charged per month for the WHMCS license. MechanicWeb also offers custom Reseller plans with child accounts with 2+ CPUs and 2+ GB RAM.
  • Brixly Reseller Pro plan offers unlimited storage, 100 cPanel accounts, disk I/O speed of 50 MB/s, 2 GB of RAM, and 2 CPUs. They have the same server speed as NameHero. Brixly offers the option of upgrading specific child accounts to 4 GB RAM and 4 CPUs for just $7 per month. Such resource allocation can be beneficial during the holiday season and matches the top-tier Business Cloud shared hosting plan on NameHero. An additional $10.40 is charged per month for the WHMCS license.
It would be helpful if you were aware that the reseller plans presented above offer a much more efficient software stack based on LiteSpeed Web Server Enterprise, which also provides full protection against “brute force” login attacks. Additionally, the hosting providers offer malware scanning and web application firewalls (WAF) – imunify360/ModSecurity at NameHero, imunifyAV+/CSF WAF at MechanicWeb, and BitNinja Security Firewall and BitNinja SiteProtection with real-time malware removal at Brixly. Cloudways does not offer such protection. Their offer is to spend $16.5 per month per domain for Sucuri – that’s an extra $1.65k per month!
Each of about 75-100 child accounts on the reseller plan will be capable of serving about the same number of visitors compared to the entire Cloudways plan. Any person recommending that multiple domains are best hosted on the shared vCPU Cloudways plans is missing the necessary skills to analyse different hosting options or is a pure scammer driven by the affiliate fees from Cloudways.
Short NameHero appriasal
If you can choose the Michigan or Kansas location, you will benefit from using NameHero’s technical support. You should select Brixly for UK clients. If your customers rather appreciate a 90% faster CPU performance, you should consider MechanicWeb servers in Los Angeles, Dallas, and Germany. They offer a similar 90% boost to the best CPU speed on Vultr HF, resulting in a faster server response to website visitors. Faster websites make for better conversion, so consider this option to make your customers happier.
LiteSpeed HostMantis servers in OVH data centres in Australia and Singapore can serve South Asia. OVH’s reputation had suffered after the fire at their Strasbourg Data Center in March 2021, so some of you may be wary of OVH when deciding to host with them. HostMantis stores all backups with backup providers not affiliated with OVH as an additional layer of redundancy. In the event of an OVH data centre failure, HostMantis would immediately restore any server from the backup to another Data Centre. HostMantis offers 200 GB of storage, 150 cPanels accounts, a free Blesta licence, 100 MB/s disk I/O speed, and 1 CPU/ 3 GB RAM per child account. As a bonus, HostMantis provides a Redis object cache to reduce the load on your CPU when serving requests for dynamic HTML files.
OVH provides a mix of fast servers such as Intel-Xeon E-2136 and a further 10% slower E3-1270 v6. The main drawback of this option compared to the MechanicWeb offering is significantly slower CPU performance. Various benchmarks suggest 60% slower servers at HostMantis. Please refer to exact details in our benchmark tables under EthernetServers entries (one of the major Resellers of HostMantis). You should also be aware of a traditional restrictive HostMantis fair use policy of 25%, which we estimate would result in serving about four times fewer visitors per child account compared to MechanicWeb.
Whenever you deal with HostMantis, be cautious. Check out the complaints of customers whose services were terminated on a whim because they dared to ask for some extra technical support. The company caters to experienced users who don’t need “hand-holding”. Avoid HostMantis if you know you’ll be a high-maintenance user! Expect that a one-person micro-company will occasionally display an awkward communication style, and you’ll avoid unpleasant surprises.
In addition, we can draw on our own experience testing Enterprise Entry plans with 2.5 CPUs from HostMantis. Although 40 Entry Processes were allocated under cPanel, our tests show that this account can only handle 5 NPROC, an overall number of processing units available inside LVE. The K6 load test was unable to use more than 1.66 CPU cores. The other hosts we tested before never exhibited this behaviour. We created a technical ticket describing the problem and received a response stating that our account was terminated and a refund was issued. The way Scott Newton, the sole proprietor of HostMantis, treated us was shocking – we did not mention that our account was an empty shell used to test his setup.
In case your customers’ visitors are from India, Pakistan or Bahrain, ExonHost servers in Bangladesh will serve them the best TTFB values. They offer an E3-1270 v6 server, similar to the one HostMantis uses in the OVH data center in Singapore. With ExonHost, you get 50 GB of NVMe SSD storage, 50 cPanel accounts, and 1 GB RAM / 1 Core CPU per child account.
So, the web agencies can use white-labelled hosting for their clients at:
  • UK (Brixly) to server customers in the UK
  • Germany (MechanicWeb) to serve customers in the UK,  Europe, Turkey, Israel, and India.
  • Los Angles (MechanicWeb) to serve clients in Japan, South Korea, and the West Coast.
  • Dallas (MechanicWeb) to serve customers in the South Central US and Latin America.
  • Michigan (NameHero) to serve clients in American East and Canada.
  • Sydney (HostMantis) to serve customers in Australia and New Zealand.
  • Singapore (HostMantis) to serve clients in Malaisia, Hong Kong, Taiwan, and South Korea.
  • Bangladesh (ExonHost) to serve clients in India, Pakistan, and Bahrain.

Dedicated managed VPS hosting

If you are asked to move your site on a virtual private server (VPS), be aware that the monthly fee will substantially increase. You can expect to pay $120 per month for this service, including the fee for the LiteSpeed Web Server Enterprise Site Owner plan  (limited to 8GB RAM on the virtual machine), cPanel and WebHost Manager (WHM), external backup, and a bunch of other software tools like Softaculous, JetBackup, and imunify360 you get for granted at the shared hosting. Instead of MailChannels, we recommend upgrading with SendGrid services as they offer better value for money. And last but not least, paying a higher bill for VPS means you can rest assured that your Redis installation will be running in a fully secured environment.
For those looking for an affordable upgrade from the semi-dedicated web hosting, NameHero offers excellent performance with a fully managed Hero 8 GB Cloud plan. And it also comes very reasonably priced at $45 per month for the first year and $88 per month after that. With these VPS plans, CPU resources are not yet dedicated, but the RAM and disk space is fully dedicated.
Good VPS providers do not oversubscribe shared CPU, but their VPS plans are priced accordingly to how many users are on the server. For example, NameHero monitors the average CPU usage on each server to keep it below 70% to allow for spikes in CPU usage. That’s exactly how they monitor CPU consumption for shared hosting plans. Since allocated RAM resources are dedicated, the number of users per server is automatically reduced in VPS plans compared to shared hosting. Before you start asking whether the CPU is fully dedicated, you should expect to find four times as much RAM in GB as the number of CPU cores. The majority of providers understand that average users need large CPU spikes, but that their average consumption is not as high. Therefore, such users benefit from the VPS plans with a several times larger but shared consumption of CPU resources.
Before you subscribe to NameHero’s VPS, you might want to consider MechanicWeb’s Elite plan for $53 per month for the first year and $71 per month after that. This Semi Dedicated plan is hosted on nearly twice as fast hardware and offers the same number of 8 CPU cores, 50% more RAM and twice as much SSD storage. This Semi Dedicated plan is an interesting and more affordable alternative to NameHero’s VPS hosting, giving you a much more powerful website if you decide to stay with MechanicWeb.
With MechanicWeb’s Semi Dedicated Hosting, you get VPS-like performance for a much lower cost since you don’t have to pay for licenses or manage the server. Shared VPS and Semi Dedicated Hosting both offer scalability. Upgrading the shared VPS require a reboot. Semi Dedicated Hosting can be upgraded seamlessly.
Root access is the main difference between Semi Dedicated Hosting and shared VPS. You can use root access to make changes to the shared VPS’s configuration and to install/uninstall specific software, making shared VPS plans more customizable. Semi Dedicated Hosting only supports basic tasks that do not require root access, for example, changing PHP versions, enabling or disabling PHP extensions, and installing WordPress.
Consider MechanicWeb‘s fully managed shared VPS plans if you need root access. The HF-68 plan offers 6 CPUs and 8 GB of RAM for $51 per month for the first month and $81 per month after that. Considering that the server speed is twice as fast, this MechanicWeb shared VPS plan offers much better value than similar shared VPS plans from NameHero. MechanicWeb offers broad geographic coverage with servers in Dallas, Los Angeles and Germany.
MechanicWeb’s shared VPS service is based on KVM (Kernel Based Virtual Machines), a true virtualization technology. KVM VPS is a virtual machine, similar to a small dedicated server. Any software can be installed on it as with any dedicated server. Shared VPS’s require their own set of software; therefore, license fees must be paid. KVM VPS accounts offer dedicated disk space and RAM by default, threfore they offer better isolation than CageFS and are less likely to be affected by bad neighbours.
OpenVZ, another virtualisation technology for VPS plans, shares RAM and storage among multiple accounts. OpenVZ offers very similar isolation between VPS accounts to CageFS for shared hosting accounts. OpenVZ provides easy possibility of overselling resources on VPS accounts. From the point of view of resource allocation, there is not much difference between VPS hosting using OpenVZ and ordinary shared hosting. A KVM VPS hosting provider makes more sense if you are thinking about switching to a VPS account.
Dedicated CPU VPS plans are much more expensive, but are still available if you need them. These VPS plans are the closest thing to a small dedicated server, created as a fraction of a larger physical server. These plans are appropriate only if you prefer to have a CPU resource that is four times smaller but is exclusively yours. Another, more appropriate use case is the requirement for unusually large RAM resources to keep a large object cache for a website which demands heavy dynamic changes to its content.
MechanicWeb offers Application Hosting with dedicated CPU, RAM, and disk space allocation on Ryzen Zen 3 servers. The top tiers are 8 CPUs / 32 GB RAM plan, and 16 CPUs / 64 GB RAM  plan. The cheaper junior plan with 8 CPUs / 16 GB RAM still offers shared CPU resources. With these plans, you can easily create and start to grow social network websites similar to Facebook using the BuddyBoss WordPress platform, which also integrates Zoom meetings and webinars with a learning management system (LMS) software. MechanicWeb’s Application Hosting plans use the fastest Ryzen Zen 3 servers, offering computational power resources unmatched by any other provider on the market. The plans provide free offsite backups, LiteSpeed Web Server Enterprise, Redis, Memcached, OPcahe, Softaculous, ModSecurity, CSF, Imunify AV+, and some paid addons. There are also some limitations since it’s not a VPS.
You can keep upgrading the hardware to cover for a massive spike in the traffic during the festive season, renting a whole dedicated managed AMD Ryzen Zen 2 3900x server with 12 CPU (=24 logical CPUs) cores and 64 GB RAM from Mechanic Web. It will provide about three times more processing power than HF-68 plan but now all CPU recources are fully dedicated to your account. As a drawback, it will command an overall monthly bill approaching $400, including software licenses. For enormous traffic volume, it might be best to unload static HTML files to QUIC.cloud CDN while serving all other static content through the Geo-Replicated Perma-Cache of the BunnyCDN as described in our separate paper.
Zen 3 AMD Ryzen 9 5950x with 16 CPU cores (=32 logical CPUs) and 128 GB RAM is the most affordable super powerful server on the market right now. Its processing power is almost double that of AMD Ryzen 2 3900x server. AMD Ryzen 9 5950x are currently available from MechanicWeb in Dallas and Germany but will be more widely available in the coming months. AMD has taken a big step forward with Ryzen Zen 3. The 45% performance improvement over AMD Ryzen 2 makes them an ideal workhorse for even the most demanding projects. The GridPane team recommends them for the most advanced users.
On the dedicated server, please start with LiteSpeed Web Server Enterprise Site Owner Plus or Web Host Essential, depending on your needs. There is no direct correlation between the number of LiteSpeed Workers and the number of CPU cores, but rather between the number of PHP child processes the Workers create. The parent process spawns a child process, which delivers the web page. Each child process increases the CPU usage of the parent process. Once the CPU consumption of the parent process reaches 80%, it is time to get another worker license.
There are cases when one worker license is sufficient for a 32 CPU core server; there are cases when an 8 CPU core server might require two workers under the Web Host Professional plan – all depends on how much its parent process consumes. LiteSpeed Web Host Elite comes with an unlimited number of workers for the most demanding users.
If you need to scale your website, but even a dedicated AMD Ryzen 9 5950x server doesn’t seem like enough for the task at hand, then it’s time to start looking into further options. Your sysadmin will likely prefer to use a high-performing HTTP load balancer and application delivery controller, such as LiteSpeed Web ADC. This software allows you to scale up your WordPress site by deploying a cluster of High Availability servers. This solution is fully integrated with the LiteSpeed Web Server and provides synchronised web cache content for all cluster nodes.
Non-techies go with managed VPS hosting because it’s the easiest way to set up a super-fast host. Kind of like buying a new computer and just turning it on. Getting an unmanaged VPS is more like buying individual computer parts and putting them together yourself. Most tech-savvy owners respect the complexity involved and prefer to hire sys-admin to deal with it. Setting up a server takes like 2 hours to build out and a few minutes of maintenance every couple of months. None will be proactively maintaining your server. You call your managed host provider when you have an issue, and THAT’s when they fix things. The same goes for having your own server and sys-admin. The only difference is you save a ton of money.
Even though LiteSpeed Web ADC is software-based, it can compete with hardware load balancers. It should be possible to add as many servers as needed. Scaling up your application using this method is the most cost-effective option. Scaling up first using AMD EPYC Zen 3 servers is a more expensive alternative to LiteSpeed Web ADC.
Using AMD EPYC you have several steps to increase your server’s computational power by more than four times. All servers listed below are based on the most recent 3d generation of AMD EPYC processors introduced in Q1 2021. However, please be aware that AMD Ryzen Zen 3 is about 75% faster and is significantly cheaper to rent.  The family includes AMD EPYC 7443 (3.6 GHz turbo all cores) with 48 cores, AMD EPYC 7513 (3.2 GHz turbo all cores) with 64 cores, AMD EPYC 7643 (3.4GHz turbo all cores) with 96 cores, and the expensive monster AMD EPYC 7763 (3.2 GHz turbo all cores) server with 128 CPU cores (=256 logical CPUs) and 512 GB of RAM. All websites are different, but we believe that using the CDN services described above will allow you to serve the traffic of more than five million visitors per day or about 150 million per month.
When selecting your VPS configuration, it is important to consider that now you are responsible for controlling the load of your server. You should monitor that maximum utilisation does not exceed 50% capacity to serve a sudden spike in traffic. You also need be ahead of the game and plan for extra visitors when running ads. Terrible things can happen if your website suddenly crashes during an advertising campaign – especially since successful promotions often lead to persistent increases in visitor volume. That could overload servers with less than adequate resources available. Plan with your hosting provider ahead of time so that they’re ready when you need them.
If you don’t have a sysadmin capable to configure LiteSpeed Web ADC but have money to burn, your best option may be cloud service management platforms provided by a third party. You might want to talk to providers like GridPane, RunCloud, UpCloud, UKHost4u, or App Platform of Digital Ocean. They offer node scalings that will allow you to spread your website over multiple servers. With such service, they provide an easy UI driven control panel so that customers don’t have any worries about scaling out or adding more resources.
Please take into account that MechanicWeb’s dedicated server and shared VPS AMD Ryzen Zen 3 deals are superior to any dedicated cloud vCPU plans, as cloud CPU resources are almost twice as slow. A physical VPS or dedicated server gives you access to all CPU threads. With dedicated cloud plans, you get only one thread per CPU (called vCPU). Resource-intensive tasks will benefit from multithreading. Thus, Vultr HF and other cloud services are not the best choice for demanding applications. And even GridPane is ready to accept it by recommending to consider AMD Ryzen Zen 3 gaming servers from OVH.

The opinions and discussions

We invite active participants of numerous Facebook groups to express their views.
We are all entitled to our own opinions based on our actual experiences, and that is indisputable! Below is a checklist of things to consider when choosing a hosting company:
  • Pricing to fit your budget.
  • Speed and user experience; to get an objective balance, I read more about negative experiences than positive ones.
  • Security, and whether the hosting company takes exceptional care of it.
  • A detailed analysis to determine if your preferred plugins can run on the provider’s servers and/or if specific plugins are banned.
  • A fast and high-quality support, available 24 hours a day, seven days a week.
For someone who needs to choose their (perhaps very first) WordPress service provider, there are several things you should consider. What kind of websites are you hosting? Who are your clients? How many websites are there? Basically (because I could go on for hours) you need to think about scalability. CDNs are no longer considered suitable for businesses in Europe due to concerns about GDPR. LiteSpeed hosting provider offers true scalability for your traffic without a CDN.
As someone who likes to experiment a lot, good customer service that is available (preferably via live chat) is very important to me. I have no qualms about walking away from cPanel. I have a fast, spacious Plesk server with a Dutch hosting company because I have worked with them before and they have excellent customer service.
Hey WebWhim, that’s one massive article you’ve got going on. Impressively in-depth! It will serve you well for organic backlinks in the years to come.
The bare metal VPS with the management layer is especially good for developers and can also be used for hosting. Professionals who self-host on VPS can easily compromise security. For this reason, I use management layers on VPS, but I can still access root via SSH. My advice for management layers is to look at Gridpane. You get all the security features you could ever want.
You are definitely on the right track! LiteSpeed, when it works for someone’s specific configuration and needs, is a killer.
I am currently using GridPane, which can be described as an enterprise edition of Cloudways for WordPress agencies. As a user of GridPane, I know they have developed a LiteSpeed stack that integrates with OpenLiteSpeed (OLS). A free OLS is not as feature-rich as LiteSpeed Enterprise, the paid version used by many shared hosting providers. OLS on GridPane is in its BETA release, which means that any support ticket is treated as a bug report, with no commitment to a quick response. As soon as it leaves BETA, I’ll give it a try! Although Cloudways has discussed similar moves, no development has happened yet.
Until now, I only knew about the NitroPack article on this research portal. The more I look around, the more I realize how thoroughly this site covers every topic. Really great work!
This is a great, very in-depth article. I learned a lot more about WordPress hosting and will keep this article for reference.
I switched from SiteGround to Cloudways a few years ago… and boy oh boy I had no idea about what was written in this article. The site speed was a little slow at times and the tech support was hit or miss. I always felt that the prices for larger servers were very high, and a couple of years ago I switched to GridPane. I have had virtually no problems with my websites and am very satisfied with the GridPane platform. I am so happy with them that I have referred several of my clients. Even though the prices are not quite as reasonable now, the service is good and the support is great as well.
I have always read a lot of good things about LiteSpeed, and after reading this article, I may just bite the bullet and finally try them out.
There are a lot of affiliate reviews out there, which makes it difficult to determine if a review is genuine or if it’s a marketing scam. I am sorry for readers who do not have trusted colleagues to share stories and opinions with, and I hope this article can serve as a foundation for some of them.
My own decision depends on speed, TTFB, backups, security, and staging environment. Contact the vendor via Live Chat to test the waters. When you talk about staging, emails, restoring from backups, and estimated response time from request to speaking with a representative, what kind of feeling do you get? The cheapest quote is not always the worst idea, and the most expensive quote may not offer the most value. Ask around for feedback from freelancers and web agencies, read customer reviews, and talk to actual users of your chosen hosting provider.
In my experience, many customers waste money by opting for a more expensive plan, believing that their traffic will skyrocket. In reality, they rarely use the resources available to them under their hosting plan. Most clients use random hosting providers, and I often consolidate their domains onto a single account. Therefore, ease of migration and good support during the migration process are key decision points for me as a developer. I also consider all other aspects such as SSL, email, backups, bandwidth, security, and technical support.
Anyone looking for hosting advice in Facebook groups should only read the responses that provide a credible explanation. You might need to try several hosting providers to find the one that works best for you. Don’t be afraid to look beyond your own country. I often hear that hosting the site in another country hurts rankings, but I’ve yet to observe the proof.
Once you find a good hosting provider, stick with them. Make sure you have the option to upgrade your hosting service later. New websites can quickly grow in size and traffic, and you don’t know how soon your client will need to upgrade their hosting plan. Depending on the client’s budget, they may use an advertising agency or contract with an aggressive SEO firm. A website may suddenly receive a flood of visitors. The structure of your hosting provider’s offerings needs to accommodate this, regardless of what type of plan you initially chose.
The article states the facts upfront. As a Malaysian web design agency, we are always considerate towards the client’s budget for hosting as it is a recurring expense. Usually, a shared hosting plan is recommended so that the client can upgrade gradually as needed. To ensure reliability, scalability and security, we select responsible hosting companies. Do not just blindly follow recommendations from affiliate communities. Check the server specifications offered to you and determine if they offer value for money.
I firmly believe that LiteSpeed is the best web server on the market, especially if it has NVMe SSDs. It is a good combination for shared hosting.
A good hosting service is essential, but your website needs serious updating for new Core Web Vitals. Upskilling and learning the latest best practices is a big challenge for all developers and designers. Consider the best UI/UX, optimizations, caching and CDN. Remember to constantly tailor your content to the viewers, not just the Google bots.
In France, most SMEs do not know much about technical benchmarks and are unlikely to ask for them. The companies focus primarily on low-cost localised shared hosting in France (also for GDPR reasons) and French-speaking support staff. The quality of the support staff and their responsiveness are essential. As a result of Core Web Vitals, everyone is asking for speed but is unwilling to go beyond shared hosting for financial reasons. We also consider ease of use of access panels (UX might be a problem with some hosting), but we usually do not choose to host using providers from other EU countries.
A successful website depends on a good hosting provider. Choosing a reliable hosting provider within your budget will give you peace of mind. Most people will hesitate to pay more than $15 per month, but others will spend more than $100 for a robust setup. Because of Google’s new Page Experience metrics, you should also find out if your host will be fast. Unfortunately, we are often in the dark when it comes to evaluating speed.
I have not personally used most of the hosting companies listed in this article. However, I am impressed with their extensive and detailed coverage in the study. Those companies will be a good choice for many businesses looking to switch to a fast and reliable hosting service. This article covers everything in great detail and gives you the opportunity to decide for yourself what to choose.
Choosing the right hosting provider for your website is tricky and something that should not be taken lightly. Many people decide with their wallets, which is not the right way to go. I can give you many examples of expensive hosting providers with crap offerings and many cheap ones with excellent deals. Also, remember that a decision is not forever. Moving a website to another provider takes skill but is relatively easy. So do not get stuck with a lousy hosting provider and suffer from poor website performance, dropping visitors and bad conversion. Reading this article will give you good pointers. There is no one size fits all. Decide for yourself what is important, get advice from professionals and then test and optimise yourself.
I suggest that you should also keep in mind the quality of customer services. Technical performance is secondary. I am perfectly content to not have the best provider and to accept their technical shortcomings (to a certain extent), as long as I know they support me and I can rely on them. If you are just starting out with web design, this is even more important. You will then have to rely a lot on the hosting company as everything is new to you. If something goes wrong, you should feel that your provider has your back.
Based on my own experience, I can fully support the article’s focus on LiteSpeed servers. Also, I fully agree that any recommendations should be ignored if they are followed by an affiliate link. As for me, I prefer to pay attention to the recommendations of genuine users. Visit the website. If it looks like the offer is decent, keep looking on the web and check other information (good and bad). Again, ignore comments from people affiliated with the company. The same goes for sponsored ads, sponsored videos, etc.
Hosting is an essential component of the success of online business. Think of hosting as a business partner. Several factors should be considered when choosing a hosting partner:
  • The servers should be located as close to the target audience as possible. You should not use hosting servers located in the US if your target audience lives in Europe.
  • Security is a crucial aspect of digital business. You should always ask your hosting provider how they will secure your website(s). Have them describe their tools and tactics to you.
  • Uptime is an important factor. There is no such thing as 100%. There may be some downtime, but it should not be too long.
  • Support 24×7. Support is what makes or breaks a hosting solution.
  • Budget. Many people do not spend enough money and compromise on the above. This is not a good idea!
  • Consider a scalable solution.
I am definitely in favour of switching to a LiteSpeed Server. You will notice the difference when you make the switch. In the past, I have used pretty expensive managed private servers that were not even as good as inexpensive shared hosting on LiteSpeed servers. But finding speedy servers is just a first part of selecting your hosting provider. As for the price, there are usually lots of options if you are looking for discounts, especially on Black Friday.
However, still, choosing your provider carefully has the biggest impact. Ideally, you want to get 24/7 support and should have options for scaling in the future. But, the deal breaker for me is solid & advanced tech support available on chat, especially if you are not familiar with the very technical back-end settings. Having tech support that will jump right into the code to fix things is incredibly important. I have used many hosting providers, and some of them are great at this, while others are terrible. NameHero, my current provider, is fantastic with this. A reseller account with them is affordable, and the free transferring for multiple websites was really fast and they handled all of it, including performing quality checks.
If you have your own back-end developers, all of this may not be necessary. However, for the rest of us, advanced tech support available 24/7 on chat is absolutely essential.

Next steps

There are many hosting providers – each with its own advantages and disadvantages – but it can be hard to decide which is the best one. We have researched and sorted out the most appropriate hosting providers who offer all of the services discussed in this article’s first section. Our separate report has covered in detail some of the providers you will have to choose from. The obvious winner is MechanicWeb who provide their plans at the most affordable price using fastest and most optimised servers.
If you’d rather pay more for marketing than for speed, consider NameHero. Although their servers are twice slower, their plans still offer one of the best deals by giving you more CPUs. The company has also been named a 2020 Best Web Hosting Company winner. They provide excellent service at an affordable price – so why not study their Turbo Cloud hosting and secure the best offer for less than $8 per month? They keep their hardware in a top-level DataCenter that is managed by Liquid Web. About 500,000 websites are hosting with NameHero and its success to receive an award shows a large user base that’s ready to vote for its champion. NameHero has an extensive library of videos from CEO Ryan Gray explaining how to use their services.
Discout at NameHero
When websites attract heavy traffic, they overwhelm the provider’s hosting capacity. Adding a CDN service is an excellent way to get your website in front of more people and grow your audience. If you have a website with high traffic, then we recommend giving BunnyCDN a try. It will store a permanent cache of your website’s static files like CSS, JS, and images in its Geo-Replicated Perma-Cache servers. These servers will provide your files to the cache of 30 servers located around the world, shielding your origin from traffic. Your origin server will be left with the task of delivering HTML files to visitors, reducing the bandwidth load at your origin by a minimum of thirty times. BunnyCDN services cost you about $1 per month. It is a much better option than upgrading the level of your subscription with your original hosting provider.
If your website is hosted on a LiteSpeed Web server, you can reduce your workload by serving HTML pages through QUIC.cloud CDN. This CDN gives 10 GB of free traffic per month, leaving your origin server to only serve dynamic HTML files for your eCommerce clients. Please read our blog post “How to Choose Your Best CDN Provider” for more in-depth technical information about the services discussed above.
Testimonial
Web design agencies are essential for the success of any website. Selecting a proper hosting and content distribution network provider can help improve the page loading time but to succeed, you need a professional web design. Website design tools are increasingly commoditised, leading to an increase in the number of web designers with limited technical knowledge. Many of these designers cannot guarantee high-speed page loads for mobile devices or desktops. Our article “How to Choose a Web Design Agency” includes clear criteria for evaluating web design companies.
We’re one of the best web design agencies in Cambridge, and we specialise in speed optimisation. Our company has been in the graphic design industry for years, with over seventeen years of experience. We work hand in hand with our customers to understand their needs and create a strategy exclusively for them. We provide a comprehensive design process combined with an affordable web design package to all our clients. Please visit our website’s page about our web designs services for more information on what we offer.
Through our comprehensive solutions, we can help your business from logo designs to designing eCommerce stores. We provide a range of services, including technical SEO and relocation assistance to better web hosting. With our graphic design expertise, we are prepared to meet any other creative need that may arise.
Please contact us at svetlana.zh@webwhim.co.uk if you need any assistance, and we will be happy to help you.
Visit our WordPress research portal for more information. We spoke with plugin developers and hosting providers to compile six articles about how to become a better WordPress user. A knowledgeable web design agency will make all the difference in the speed of your site. You also need a set of plugins, tools, and a good hosting provider and a fast CDN to keep up with incoming traffic. If you don’t do these basics, your website will not pass the Page Experience signals introduced by Google.