I used to wonder how hosting companies (you’ll easily find some examples online) could offer ‘reliable, fast and secure’ shared hosting for £2 per month and make a profit.
I’ve use several hosting setups in the past to run websites built using WordPress, React and basic HTML and CSS. I’ve also helped diagnose issues on several shared hosting setups.
As I’ve started to build my own shared hosting setup for my client sites, I’ve come to understand how these large hosting companies operate.
How much it costs me to host 15 WordPress sites
You can rent virtual private servers (VPS) from hundreds of companies, including Microsoft Azure, Amazon Web Services and Digital Ocean. You’d normally host a website on a VPS if you wanted dedicated resources to provide a reliable speed.
I currently rent virtual servers from Digital Ocean, which they call Droplets. The smallest CPU-optimised Droplet can be rented for about £35 per month. This Droplet has 2 dedicated virtual CPUs, 4GB of RAM and a 25GB SSD. This is the minimum Droplet spec I’d recommend for running a WordPress website. You’ll find cheaper VPS setups elsewhere, but I wanted a CPU optimised server which could be managed by SpinupWP.
SpinupWP, who make it easy to run WordPress sites on Digital Ocean Droplets, say it should be possible to run between 10 and 20 websites comfortably on my spec of server. Here’s their guide on picking a suitable size server: https://spinupwp.com/doc/which-server-size/#upgrade-my-server
Over the past 14 days, I’ve run 15 WordPress sites on that Droplet. It has cost me about £2.33 per site this month.
The resources used by my 15 sites
15 small, moderate traffic sites are running very well on a 2 vCPU, 4GB RAM, 25GB SSD server.
It’s worth pointing out that these sites don’t run eCommerce stores, or eLearning courses – both of which can be resource intensive.
Some of these sites are just demo, proof-of-concept and staging sites which don’t receive traffic.
CPU usage
I’ve just pulled this graph from Digital Ocean. It shows how my CPU load has varied over these past 14 days, whilst I’ve been running 15 WordPress sites on 2 vCPUs.
The biggest spikes peak at 40%. I’ve never seen a spike above 50% since moving to this VM.
Memory usage
The memory graph shows how the 15 sites have shared 4GB of memory over 14 days. Again, the highest spikes peak at <60%.
The issue with most shared hosts
A lot of shared hosting companies rent VPS servers, just like those Digital Ocean Droplets mentioned above. They’ll then cram as many WordPress sites onto one server as possible, to reduce their cost per site and make a profit.
I know that 15 websites run very well on my £35/month server. I could probably run 20-30 websites on that server, but if a couple of sites receive a spike of traffic, it’ll risk slowing down other sites. I don’t appreciate unpredictable website loading speeds, and I don’t think my clients do either.
In order to sell hosting for £2 per month, the cost per site has to be incredibly low. Most shared hosting companies are therefore gambling on websites receiving low amounts of traffic.
The benefit of using my hosting over generic shared hosting
I have far more control and visibility of what’s going on, by renting a virtual private server. I know that my client sites and staging sites are running as efficiently as possible.
I’ll get alerts before my server runs out of resource or if a spike exceeds 80% of resource. This means I can be confident that my websites are consistently running quickly. If a website starts hogging resources, I can move it to a new server with ease.
Would a shared host tell you if their overpopulated server was causing your website to run slowly at peak times?
By having full control over my hosting setup, I can provide consistant speeds to my clients. I’m also aware of how each site is built, what it stores and what visitors might use it for – therefore making my setup more secure than generic shared hosting.
How companies are providing hosting for £2 per month
To conclude, companies offer very cheap WordPress hosting by cramming your website onto a full server. They’ll use clever terms & conditions to protect themselves against your unpredictable loading speeds.
