The Machines Behind the Machines

Hi, Paul here again, CEO of CreeperHost.

We’ve already talked about the hardware we build and the network that connects it all.
Now it’s time to look at what runs behind the scenes, the software, automation, and glue that keeps everything working the way it should.
Every control surface, API, and background process that makes your server tick was written by us.
No reskins, no off-the-shelf panels, just years of development and rebuilding until things worked the way we wanted.

Built Before Containers Were Cool

Every customer server we host runs inside an isolated container environment. It isn’t a full virtual machine, but it isn’t a simple sandbox either.
We’ve used this setup since 2011, long before containers became a buzzword. Back then it was just the most practical way to run hundreds of servers efficiently with proper isolation and near-native performance.

Our storage system is designed for speed and consistency. Each container has its own virtual drive, which behaves just like a real hard disk instead of sharing space with everyone else.
That means smoother performance, predictable write speeds, and no shared file limits getting in the way.
Every server gets its own environment, so game saves, logs, and configs behave exactly as they would on a dedicated machine.

For backups, we built Cloud Saves, our own S3-compatible backup platform that runs entirely on our own infrastructure.
It automatically keeps versioned copies of your data, so if a mod update goes wrong or a world gets corrupted, you can roll back with a few clicks inside CreeperPanel.

All backup data stays here in the UK, in the same server room as our core systems.
It can be a little slower than using a big cloud provider, but I’ve never liked the idea of handing something that important to someone else.
Nobody out there is going to care more about your data than we do, how could they?

The Core Platform - Aries

Our main API, Aries, has been around since 2013. It started in PHP and has evolved through a lot of rewrites and upgrades over the years.
These days, Aries is the backbone of everything once a container exists. It talks directly to customer containers, handling management tasks like installing games, creating new instances, updating configurations, and keeping everything running smoothly.

While newer systems like Hermes, Demeter, and Hera now handle many of those tasks faster and more efficiently, Aries can still do the job if they’re ever offline.
It’s capable of everything, from file access and process management to command execution and status reporting, just not as quickly.
It’s the system that ties everything together, the part that keeps the rest of the platform in sync.

Speeding It Up - Hermes and HermesMux

To make things quicker, we built Hermes, a communication daemon designed to replace slow SSH2 handshakes with a much faster, lightweight authentication process.
It made every action between containers and the control API far more responsive, especially when handling lots of smaller commands.

We later added HermesMux, which manages connections between Aries and Hermes inside our Kubernetes cluster.
HermesMux keeps those links open persistently, allowing data to move instantly between containers and control systems.
The result is near real-time communication between CreeperPanel and customer servers, even across long distances.

Scaling Up - Our Kubernetes Cluster

As we expanded, we moved Aries from separate regional servers into our own self-hosted Kubernetes cluster.
It’s a mix of different processor types(x86 and arm64) and automatically scales up or down depending on demand.

We run dedicated control planes and dedicated storage systems for reliability, plus a range of worker machines for handling everyday tasks.
In total, the cluster has around 1.4TB of memory and 537 processor cores.
It isn’t rented or hosted elsewhere; it’s our hardware, in our racks, managed by our own team.

It’s worth noting that this cluster doesn't run game servers; it runs all the control systems that manage them.
That separation keeps things stable, scalable, and much easier to maintain without impacting anyone’s gameplay.

Everything runs inside a private network with a single managed entry point.
That setup keeps things fast, consistent, and isolated from anything that shouldn’t be talking to it.

Smarter Access - Demeter and Hera

As Aries grew, we wanted faster ways to handle common tasks like file editing or process checks without pushing everything through one big system.
That’s where Demeter and Hera came in.

Demeter handles file access using a web-based system that works a bit like a streamlined file manager.
Actions like editing configs, accepting the EULA, or checking logs can now happen instantly.

Hera looks after process monitoring, showing what’s running inside each container in real time.
Together, they take some load off Aries and make CreeperPanel feel quicker and more responsive for customers.

Making Games Run Anywhere

Most of the games we host run directly on the operating system for maximum performance. But some games need a specific set of libraries or quirks to behave properly on Linux.

To make sure everything works the way it should, we use Steam's Pressure Vessel, Steam Runtime, and Proton, the same systems Valve uses for SteamOS and the Steam Deck.
We often build and package these ourselves so they fit neatly into our setup and behave the same way across all our servers.

For Windows-only games, Proton gives full compatibility without the overhead of running a virtual machine. It’s efficient, reliable, and benefits from years of work by Valve and the Linux gaming community.

Under the hood, we've made a few small improvements of our own to keep things smooth.
We’ve written some kernel-level tweaks and built a system that helps our servers safely share unused memory between games.
It’s invisible to players, but it helps us hit that no more than 60% utilisation target we talked about in the hardware post and keeps everything running consistently.

And when it comes to actually getting games running, we do the work ourselves.
We don't download someone else's scripts or community templates and hope for the best.
Every title we support has been installed, tested, and tuned by us to make sure it works properly from the start.

Provisioning: Containers All The Way Down

When someone places an order, it doesn't just appear instantly by magic, even if it looks that way from the outside.
Behind the scenes, there’s a whole chain of systems that spin into action automatically.

Every order joins a queue watched by our own Kubernetes operator.
That operator spawns a temporary provisioning pod, which checks for available space across our fleet.
It then talks directly to the target host through our in-house orchestration system, using our own HTTP/3-based API to create the container, install the base system, assign an IP, and set up networking.

Once the new container exists, the provisioning system checks that everything is online and reachable.
Then Aries takes over, connecting directly to the new container to install the customer’s chosen game setup and configuration.
Finally, once everything checks out, WHMCS sends the customer their welcome email.

It sounds complex, but it all happens in seconds.
A container in Kubernetes spins up another container that tells a third system to make a container somewhere else.
It’s containers all the way down, and somehow, it all just works.

Daily Backups and Node Protection

Every node in our network is backed up daily using a system called Borg.
We only back up the important bits, customer data and configuration files, not the entire operating system.
If a node ever fails, it's faster and safer for us to rebuild it fresh and restore the customer data, rather than cloning everything wholesale.
That approach keeps downtime short and avoids carrying over any underlying issues from a failed machine.

Billing, the One Exception

We build nearly everything ourselves, but not the billing system.
People trust us with real money, and that's not something to take risks with.
We use WHMCS for that part because it’s audited and built for handling payments securely.
It’s the one area where using a proven system is simply the right call.

A Host That's Actually Ours

Everything you've read here was built, tested, and maintained by our own team.
No resellers, no third-party control panels, no waiting for someone else to fix a bug.
When something needs improving, we build it. When something breaks, we fix it.

We even use it ourselves. Many of our staff rent game servers through the same systems as everyone else.
It's the best kind of testing, real-world, daily use by the same people who build and maintain it.

This stack has been evolving for over a decade, and it’s what lets us keep improving without slowing down.
We don’t wait for patch notes.
We write our own.

What's Next

So far, we’ve covered the hardware, the network, and now the systems that hold it all together behind the scenes.
Next time, we’ll move on to the part you actually see, CreeperPanel 4 and our website.
CreeperPanel might feel modern, but like Aries, it’s been around a while. It was built on cutting-edge tech for its time, and it’s evolved through several major versions, from CreeperPanel 1 and 2 all the way to 4 today.
That story deserves its own post, and we’ll dive into how it came to be, what it runs on, and how it’s changing again soon.

- Paul (CEO, Founder, CreeperHost)