It usually begins with a simple question: Can I do this myself?
A photo album. A file share. A team wiki. Something you are currently using — from a provider who offers it conveniently, free or cheap, with a pretty interface. And then a moment when you realise: the data lives somewhere. Not here. Not with me. With someone I do not know, under conditions I cannot influence.
That moment is the beginning of self-hosting. And it changes more than you initially suspect.
What self-hosting teaches
Anyone who decides to run a service themselves enters a school. Not a theoretical one — a practical one.
Suddenly it is about networking: IP addresses, port forwarding, DNS, reverse proxies, subnets. Things that were invisible before become tangible. You learn how a packet travels from A to B, why a connection stalls, where a timeout originates.
It is about security: firewalls, encryption, certificates, updates, access rules. Whoever puts a service on the internet is responsible for it. That sharpens the eye for threats you previously only knew from the news.
And it is about service lifecycle management: updates, backups, monitoring, recovery, migration. A service is not installed and done. It lives. It ages. It must be maintained. Once you grasp this, you understand infrastructure entirely differently.
Self-hosting is the most intensive IT training there is. Because it is not simulated.
The honest truth about cost
Now it gets uncomfortable. Self-hosting is often more expensive than the cloud.
Not always, but often. A server costs electricity. Redundancy costs a second device. Backup costs storage. Running your own mail server costs nerves and time you would save with a hosted service. And then the labour hours: installation, maintenance, troubleshooting, updates. Hours someone has to pay for — with money or with free time.
Anyone who denies this is not being honest. The cloud providers did not invent their economies of scale. They exist. A data centre with hundreds of thousands of servers can deliver an email cheaper than a single server in a cellar. That is not an argument against self-hosting. It is a fact you need to know before deciding.
Why handing over data is no longer an option
But now we flip the argument. Because price is not everything.
When you entrust your data to a cloud service, you do not know where it physically resides. You do not know who has access. You do not know whether it is being analysed, sold, used as training material for an AI, or passed on to subsidiaries. The terms of service are worded so that you consent without understanding.
You do not know whether backups exist — and if so, according to what scheme. You can do nothing if the service goes down. You cannot walk to the hard drive. You cannot pull the plug and restart. You can only wait. For someone you cannot reach, in a system you do not control.
In a time when data is currency, when profiles are built, behaviour predicted, and decisions about people automated, the question of whom you trust with your data is no longer a detail. It is existential.
“I have nothing to hide” — the most dangerous misconception
There is an argument that always comes. “I have nothing to hide.” Said by people who close their curtains, keep a diary under lock and key, have a password on their phone — and still genuinely believe privacy is only for criminals.
That is wrong. And dangerously wrong.
Everyone has something to hide. Not out of shame, but because privacy is the precondition for autonomy. Whoever reveals everything is manipulable. Whoever has no secrets has no boundaries. And whoever has no boundaries has no freedom — only the illusion of it.
Data protection laws like the GDPR are not a bureaucratic exercise. They are the institutionalised recognition that data power is power over people. Whoever has the data has the interpretive authority. Whoever can predict behaviour can influence it. Whoever knows the digital footprint knows the person better than they know themselves.
Big Data only works because we collectively gave up. Because millions of people said at some point: “Ah, screw it, I’ll go to the cloud.” Each one individually, harmless. Together: a glass human. A transparent citizen, analysed by systems they do not understand, optimised for business models that do not pursue their interests.
Anyone who says they have nothing to hide should conduct their next conversation with their doctor, their accountant, or their bank in public. Nothing to hide? No. They simply have not yet understood that they have already surrendered what they should have protected.
Naming the downsides honestly
Self-hosting has disadvantages. Concealing them destroys credibility.
Global reachability. A hyperscaler — AWS, Azure, Google Cloud — operates points of presence all over the world. Delivering content simultaneously in Tokyo, São Paulo and Cape Town requires global infrastructure. Your own server in Alsdorf delivers quickly in Alsdorf. Not in Tokyo.
Performance. Content delivery networks cache content near the user. That is fast. That is efficient. That is something you can replicate yourself only with considerable effort.
Availability. Large cloud providers offer SLAs that a single operator can scarcely match. Redundant power, redundant networks, redundant data centres. It costs — but it works.
All of this is true. And all of it is why some services are better placed in the cloud than in your own cellar. Denying it categorically is ideological, not practical.
And yet: why sovereignty is non-negotiable
Now the turning point. Because despite all the drawbacks, there is a question that no speed and no SLA in the world can answer: Who has the control?
Availability matters. But availability without control is a promise someone else can break at any time. A cloud service can double its prices. It can change its terms. It can suspend your account — because of a flawed AI-driven content analysis, without warning, without recourse. It can discontinue the service. It can be acquired, and the new strategy no longer covers your use case.
The internet was designed as a decentralised medium. A network in which every node is equal, in which data can find many paths, in which there is no single point of failure. This architecture was the condition for the internet’s growth. It was the condition for its resilience.
Large cloud providers break this concept. They centralise what was meant to be decentralised. They build silos that barely talk to each other. They create dependencies that contradict the spirit of the internet.
When the cloud fails
And then there is the argument least often mentioned but hardest to refute: concentration risk.
Cloud providers are highly available. Until they are not. When a hyperscaler goes down — and it has happened, repeatedly — it is not one service that fails. Hundreds, thousands fail. Simultaneously. Globally. Airlines, banks, news sites, hospital IT, payment systems. All on the same platform. All affected by the same incident. All powerless.
That is the flip side of centralisation. A single entity that is reliable until it is not — and then drags everything down with it. Decentralised structures are slower, less efficient, less perfect. But when one node fails, the others do not fall with it. That is not weakness. That is resilience.
The oligopolistic prison
Let us talk about the prisons we built for ourselves.
Two mobile platforms. Three cloud providers. A handful of search engines. One social network that absorbed all the others. We live in a digital world dominated by a handful of companies — and call it choice.
It is the choice between pest and cholera. iOS or Android. AWS or Azure. Google or — well, what exactly? The illusion of options is the strongest instrument of oligopolies. You may choose — between two options that both follow the same model: collect data, create dependency, make exit difficult.
Choosing iOS means choosing a walled garden. Choosing Android means choosing a different walled garden — in a different colour. In both cases, you surrender. In both cases, the vendor decides what runs on the device, which apps are permitted, which data flows. The user picks the device. The vendor picks the terms.
This is no conspiracy theory. It is a business model. And it works because exiting is more expensive than staying. Vendor lock-in, as we know it — only this time not with a single software, but an entire ecosystem.
The way out of the prison starts small. It starts with one service you host yourself. One first step back toward sovereignty. Not everything has to be self-hosted at once. But every service pulled out of the oligopoly is a piece of reclaimed freedom.
Self-hosting — for others too
And here a responsibility comes into play that is often overlooked.
Those who have the ability to self-host — technically, temporally, mentally — bear a responsibility that extends beyond themselves. Not everyone can run a mail server. Not everyone can administer Nextcloud. Not everyone wants to. But everyone has data. And everyone deserves a place where that data is not sold to the next corporation.
Those who self-host should not host only for themselves. But also for friends. For family. For the neighbour who does not want to hand their photos to Google but has no clue about Linux. For the parents looking for a secure address book. For the club that needs a wiki it owns.
This is not charity. It is resistance. Every time someone diverts their data stream out of the oligopoly and onto a self-hosted service, the monopoly weakens. Not noticeably. Not immediately. But in principle.
And it is education. Whoever uses a self-hosted service — even without administering it — learns that alternatives exist. That data does not inevitably have to live with Google. That you can trust someone you know instead of someone you do not. That is the seed from which, one day, something grows.
Further threads worth following
Anyone engaging with self-hosting stumbles upon questions that run deep:
Encryption and key management. Encrypting your data — but leaving the key with the cloud provider — means encrypting for the provider, not for yourself. Zero-knowledge architectures are the consequence. They are laborious. But they are the only way a provider truly cannot read anything.
Backups and recovery. A backup you have never tested is not a backup. Self-hosters test. Outsourcers trust. The difference becomes visible only when it burns.
Law and compliance. Where do the data live? GDPR, server location, subcontractor chains. Self-hosters know. Outsourcers hope.
Migration. The path into the cloud is easy. The path out is not. Export formats, proprietary data formats, missing APIs — the familiar vendor lock-in pattern. Self-hosters own their data in formats they understand. Outsourcers own a subscription.
How libcom.de can help
Here it gets concrete. Self-hosting is not an end in itself. It is a means to sovereignty. And sovereignty demands know-how.
We at libcom.de have been operating self-hosted infrastructures for more than two decades. We know what it costs — in time, money and nerves. But we also know what it is worth.
Our work does not begin with a product but with a question: What do you need, and what do you want to keep in your own hands? Sometimes the answer is: a complete on-premise solution. Sometimes: a hybrid setup where sensitive data is self-hosted and resource-intensive services run where economies of scale apply. Sometimes: simply an honest assessment of whether self-hosting makes sense for your case.
We plan, implement, document and hand over. We train teams so they can understand and continue what we build. And we do not remain the bottleneck — those who leave us can keep the system running because it is documented, standardised and comprehensible.
If you wonder whether self-hosting is an option for you, or if you already self-host and need support: write to us at contact@libcom.de. We take an honest inventory — without sales pressure, with a view to what your infrastructure needs in the long term.
Self-hosting is not the cheaper path. It is often the more expensive one. But it is the path on which you understand what you operate. On which you learn how things connect. On which you keep control — over data, over availability, over the question of who actually has access.
The internet was conceived as decentralised. Perhaps it is time to reclaim that idea.