CalorieStar had been running perfectly happily on Hetzner; a uk vps comparison could help evaluate options.
The server was reliable, performance was good, and I wasn’t actively looking to move. Then the pricing changed significantly enough that it made me stop and ask a fairly simple question:
If I were choosing a VPS for CalorieStar today, would I still choose the same one?
Once I started looking properly, the answer wasn’t obvious.
That eventually led to two things: moving CalorieStar to a new UK-based VPS provider, and building a VPS comparison table because comparing hosting plans turned out to be far more awkward than it should be.
Why I started looking elsewhere
There wasn’t a dramatic failure or falling-out with Hetzner.
CalorieStar had been running well and I was happy with the service. The trigger was simply cost. A significant price increase meant it was worth taking another look at the market rather than automatically accepting the new price.
At the same time, CalorieStar itself had moved on.
The site is primarily aimed at UK users, so if I was going to reconsider the infrastructure anyway, I also wanted to see whether moving the application closer to its users made sense.
Hosting location isn’t the only thing that determines website performance. Network routing, server performance, caching, application design and plenty of other factors matter too. Germany isn’t exactly a high-latency location for UK visitors.
But if I was starting again, a UK data centre felt like the natural choice.
So I started looking.
Comparing VPS providers is surprisingly difficult
At first, I approached it in the same way I normally would.
Open a few browser tabs, find plans with roughly the right specification, compare the prices and make a decision.
That works until you have more than a handful of providers.
One company advertises:
4 vCPU, 8 GB RAM and 160 GB NVMe for £20 per month.
Another offers:
6 cores, 16 GB RAM and 200 GB SSD for £24.99.
Another quotes annually.
Another excludes VAT.
Another includes 20 TB of transfer while somebody else describes it as unlimited.
Some include IPv4 addresses, some charge separately for them, and some bundle backups while others make them an optional extra.
Then there are differences in CPU generation, virtualisation technology, network speeds, storage type and data-centre location.
Before long, I had far too many browser tabs open and no sensible way of answering the question I actually cared about:
Which VPS gives me the best fit for CalorieStar at a sensible price?
Normally, this is where I would open Excel
My usual solution to this sort of problem is a spreadsheet.
Create a row for every hosting plan and columns for the things I care about:
- Provider
- Price
- vCPU
- RAM
- Storage
- Storage type
- Data transfer
- Network speed
- Data-centre location
- Backups
- IPv4
- VAT
- Billing period
Then start filling it in.
It works.
But as I was doing this, it occurred to me that there wasn’t really anything about this information that needed to live in an Excel file on my computer.
I was manually collecting publicly available information from hosting companies and arranging it into a useful format.
If it was useful to me, there was a reasonable chance it would be useful to somebody else looking for a VPS too.
So instead of another spreadsheet, I built a simple web-based comparison table.
Building the VPS comparison table
The first version wasn’t intended to become a major CalorieStar feature.
It was simply a better version of the spreadsheet I was about to create.
Each hosting plan could be entered into the table with a consistent set of attributes, making it much easier to compare like with like.
For each plan, I wanted to capture things such as:
| Item | Why it matters |
|---|---|
| Provider | Who actually supplies the service |
| Location | Whether the server is hosted in the UK or elsewhere |
| vCPU | The compute available to the VPS |
| RAM | Particularly important for databases and caching |
| Storage | The amount of disk space included |
| Storage type | SSD and NVMe can make a substantial difference |
| Transfer | How much outbound traffic is included |
| Port speed | The network connection available |
| IPv4 | Whether an address is included or charged separately |
| Backups | Whether backups are included |
| Monthly price | The headline cost |
| VAT | Whether the advertised price includes it |
| Billing period | Monthly, annual or another commitment |
Once the data is structured, it also becomes possible to calculate things that aren’t normally shown on hosting-provider websites.
For example:
Cost per GB of RAM
Cost per vCPU
Effective monthly cost
Those numbers don’t determine which VPS is best, but they make it much easier to spot plans that offer unusually good or poor value.
Price isn’t the whole decision
One thing became obvious fairly quickly: the cheapest VPS wasn’t necessarily the one I wanted.
CalorieStar is a live application rather than a disposable development box.
That changes the decision.
I care about price, but I also care about things such as reliability, support, provider reputation, sensible upgrade paths and being able to increase resources without having to move everything again.
There is also a difference between getting an impressive specification on paper and getting a server that performs consistently in the real world.
So the table became a way to narrow down the options rather than an automated answer to the question.
A VPS with twice the RAM for another few pounds might be a much better choice.
Equally, paying significantly more for resources the application will never use doesn’t make much sense either.
Why I wanted a UK host
CalorieStar’s initial focus is the UK.
That meant I was particularly interested in providers that could host the application in a UK data centre.
There are several reasons for that, but the simplest is proximity.
For most applications, shaving a few milliseconds off network latency isn’t transformational. A badly optimised application hosted in London can still be considerably slower than a well-built application hosted in Frankfurt.
But when everything else is reasonably equal, keeping the application close to the people using it seems sensible.
There was also a less technical reason.
If I could find a UK hosting company that offered the performance, reliability and pricing I needed, I’d rather spend the money with them.
Choosing CalorieStar’s new VPS
After comparing the options, I eventually selected a new UK-based VPS provider and migrated CalorieStar across.
I wasn’t looking for the provider with the biggest numbers.
I was looking for the best balance for this particular application:
- Enough CPU capacity for PHP and background processing
- Plenty of memory for MySQL and caching
- Fast NVMe storage
- Good network connectivity
- UK hosting
- Room to grow
- Sensible monthly pricing
That’s an important distinction.
There probably isn’t a single best VPS provider.
There is only a provider and plan that best suits a particular workload, budget and set of priorities.
The VPS I chose for CalorieStar won’t necessarily be the one I would choose for a tiny WordPress blog, a database server, a game server or a large SaaS application.
The migration
Moving the server was also a useful reminder of why I prefer infrastructure that can be rebuilt rather than infrastructure that becomes a precious one-off machine.
CalorieStar uses a fairly conventional web stack, so the broad process was straightforward:
- Provision the new VPS.
- Install and configure the web stack.
- Apply the required PHP, MySQL and Nginx configuration.
- Copy the application and supporting files.
- Migrate the databases.
- Recreate scheduled jobs and supporting services.
- Test the application against the new server.
- Switch DNS across.
- Monitor logs, performance and application behaviour.
There are always little things that get missed.
Cron jobs, permissions, firewall rules, mail configuration, SSL certificates and file paths are exactly the sort of details that remind you a migration is rarely just “copy the website across”.
But overall, the move was uneventful — which is exactly what you want from an infrastructure migration.
Was moving worth it?
Yes.
The price rise was what caused me to look, but the exercise was valuable beyond simply reducing hosting costs.
It made me review what resources CalorieStar actually needed and what I wanted from its hosting as the application grows.
It also moved the application onto UK infrastructure, which better reflects the audience I’m currently building CalorieStar for.
Most importantly, it resulted in something potentially useful beyond CalorieStar itself.
The VPS comparison started as my spreadsheet
The comparison table now exists because I needed it myself.
That is usually my favourite kind of tool to build.
There was no grand product strategy behind it.
I had a problem:
I wanted to compare a lot of VPS plans without constantly jumping between provider websites and trying to remember which prices included VAT.
The obvious solution was a spreadsheet.
The slightly better solution was to make that spreadsheet into a simple web tool and share it.
I’ll continue adding providers and plans as I come across them, and I’ll update the data as pricing and specifications change.
It won’t tell you which VPS you should buy.
What it should do is make it much easier to see your options.
