Homelab: I Came, I Saw… It's Not For Me

Recent years have brought increasing popularity to self-hosting and so-called homelabs, meaning running your own servers at home and hosting the services you use on them. The reasons vary: from wanting independence from external companies and a theoretical reduction in costs (ubiquitous subscriptions), through better privacy and data protection, to simple curiosity and the desire to tackle the subject. Over the past few weeks, I was completely captivated by this topic, and after weathering many storms, I decided to share my experiences and conclusions.

Right at the beginning, I can tell you... no. As of the end of the first half of 2026: in my situation, I no longer see the point in having a homelab. I had to dedicate a lot of time and effort to reach this point. Was it worth it? Definitely yes, because it was something that interested me a lot - and in a way, still does. However, without "burning through" those few weeks, I wouldn't have the knowledge I possess now. It's a lot like testing different types of bicycles. Until I tried it, I didn't know if something was for me... or completely not for me. But let's get to the origins.

Origin, Configuration

Back in the winter, I was using a 5G internet connection that had a monthly data limit. It was never an issue, but I knew it meant some limitations: both the speed and the data cap made "homelab play" impossible, or at least severely restricted it. However, the situation changed because, since February, I have been using a symmetric fiber-optic connection, and I also have a static IP address. The decision was made quickly: it was time to try. Instead of building everything from scratch, for example by building a desktop PC, I decided to use a small laptop with good Linux support: hardware sporting a 13th-gen Intel i5, 16 GB of DDR4 RAM, and an NVMe drive means a solid power reserve alongside very low idle energy consumption. Although I tested it, I didn't go with the highly recommended and popular Proxmox. Instead, I chose... CachyOS. Arch, and a gaming edition at that, on a server? Sounds like a crazy idea? Maybe so, but for home use, it turned out to be a really good, efficient, and surprisingly trouble-free configuration. Ultimately, everything was going to run as Docker containers anyway, so does the host make that much of a difference? For me, no, and I must clearly state this here: throughout several weeks of various tests, the base system was never an issue. Of course, I could have used Ubuntu Server, I could have gone with Debian... but I like CachyOS, I use it daily, and it provides a lot of excellently pre-configured elements, so why not?

Quite quickly, a second player appeared, namely an Asustor NAS, model AS5202T, housing two 4TB enterprise-grade HDDs. I must admit that the current prices of such components are really unpleasant, and "hunting them down" was a real treat. The NAS itself also turned out to be extremely functional. I had past experiences with Synology and expected rather slow performance and limited functionality... but here was a pleasant surprise, because despite its low price, it’s truly great hardware. Under the hood is an energy-efficient, dual-core Intel Celeron clocking between 2.0 - 2.7 GHz, with a 10W TDP, allowing for hardware video decoding, including new formats (HEVC / VP9). Add to that two 2.5 GbE ports with aggregation support, and 2 GB of RAM that can be easily expanded to 8 GB - you can safely say this NAS is suited not just for file storage, but as a foundation for many services... which I managed to thoroughly test over these few weeks. The software is intuitive, offers a ton of features, and there's probably no point in writing too much about it here. Even though the hardware is a few years old, it is still supported and receives updates.

Connecting everything, configuring it, etc., took a moment, but it didn't end there - for the homelab to make some sense, to enter the world of self-hosting, I had to test some services. Initially on the laptop itself, later with the attached NAS as data storage, up to a situation where the laptop disappeared from the configuration and only the NAS remained with a rather minimalist set of services (Portainer + AdGuard Home + Caddy). I tested, I tested, and I must admit that through a process of elimination, I kept excluding more and more services from home application... Below I present a list and my observations. I actually tested a bit more, but I won't even describe minor tools like Komodo instead of Portainer, because you could say they popped in for a moment only to be quickly tossed out after another moment. Ultimately, the entire homelab was turned off and dismantled. I do not see the point in having and maintaining it.

Tested Services

Self-hosting the WordPress blog I run and which you are currently reading, dear reader: this idea seemed good, but I quickly abandoned it because it introduces a lot of overhead, primarily due to the need for backups and proper isolation of such a system. A Docker container with minimized privileges? Sounds okay, but still slightly not enough, and WordPress is frequently attacked. The second issue is exposing it to the world - opening ports in the router/forwarding them with a static IP is the easiest solution... but also the most dangerous. I did it for testing, and it quickly turned out that the router was instantly scanned, followed by automated attack attempts on Apache (I used Caddy), etc. Setting up a WAF yourself, a proper firewall? It's possible, but that means digging deeper and deeper into something that the user of the blog... doesn't see at all anyway? Instead of writing posts, I'm burying myself in technical issues, and I don't think that's the point here. You can use a Cloudflare tunnel and not open any ports, as long as we trust that provider - I use CF anyway, so I tried it. Well, it works, it's fine... but privacy advocates will say it's not for them. Considering I've had a completely free hosting account on MyDevil for several years (thanks to referral links), I decided that self-hosting the blog makes no sense.

Immich as an alternative to Google Photos: I won't hide it, this piece of software is simply brilliant work, and the authors deserve applause and donations. Quick to set up, it works, and has better sync settings than the original. But when we start digging, friction appears. I use the Google Photos search engine extensively because it simply works well. Unfortunately, on Immich, this fails in Polish; even after changing the model to the best available and running a time-consuming, resource-heavy re-indexing of everything, the results disappoint - I search for the word "sea" [morze], and as a result, I receive... shoes. Also, a photo of a dog. Well, that significantly lowers the functionality of the service in my eyes. However, that wasn't the biggest problem. Google Photos works, allowing for quick sharing among family and friends - and since a ton of people have Google accounts, it's effective. Yes, this is an example of vendor lock-in, but let's be honest: is there any point in suddenly convincing everyone that they need to use something else? Or two very similar applications? It's doomed to fail. So, the idea fell apart very quickly.

The entire Arrstack with Jellyfin, qBittorrent with a proper VPN (Gluetun), a suite of services, and automated configuration - of course, as they say, all of this is "for browsing numerous Linux images"... in very high quality, without paying any subscriptions, etc. It works, and after setting it up, it is really convenient and automatic. But... it makes sense if you actually use it. And for me, it looks like this: my loved ones and I watch movies very, very sporadically. First, we prefer to spend our time differently; second, there is simply no time for it. I’d rather go to the cinema a few times a year - for some gem movies - than constantly clog my head with "content." Series? I don't like them at all, so that fell off. I'll admit without hesitation that my TV more often plays Bluey, Paddington Bear, Kicia Kocia, or Basia... and I watch them too. Heck, it's sometimes more interesting, and definitely much more true-to-life, than many of the currently produced films.

Caddy as a reverse proxy: a simple matter, for more convenient addresses like service.lan. I have no complaints here, it is trivial to do, it works. We can use other solutions like Nginx Proxy Manager or Traefik, but overall, there's no great philosophy to it.

AdGuard Home: as a local DNS server, ad and tracking blocking. This is also a piece of good software that provides substantial capabilities. However, a problem arises: it either runs on something very stable (NAS, separate server), or in the event of any changes or minor downtime, your entire home network suffers an "outage" and lack of access to internet services. Well, that's problematic. It also opens another Pandora's box - what if I want to use this server even when I'm away from home? I have the luxury of a static IP and a router with WireGuard support, so I quickly launched a WG Tunnel on my phone: whenever I left the house, the phone automatically enabled the VPN and all traffic went through it, while DNS was handled from my local AGH. It can be done differently too: you can expose the AdGuard endpoint publicly (e.g., via the previously mentioned Cloudflared) and use DNS-over-HTTPS (DoH), including client verification, meaning without creating a public DNS, without a high risk of attack. As you can see, the matter gets complicated... I have a multi-year package for AdGuard Private DNS (the external option, with almost the same configuration capabilities) for a really very low price, so I decided it's pointless to host DNS myself. My router supports DNS-over-TLS, so all devices can use this service securely, in an encrypted manner.

Streaming music from NAS or Jellyfin: although the NAS supported such a function, it would also require the use of a tunnel. Streaming via Cloudflare is out of the question because it violates the service regulations and can end in an account ban. I tinkered, I tried... until it finally "dawned" on me - I listen to specific music, I don't have a lot of it (maybe a few GB, and likely not even that), and on my phone... 450 GB of free space is lying idle. Instead of overcomplicating things, I can upload it there, or even synchronize it using the excellent Syncthing, and have it available always, even without network access. Simple, effective... and yet it took a lot of time for me to realize it.

Monitoring stack, Grafana: this is a brilliant tool, I use it at work and really admire it for its capabilities. In the case of a homelab, it quickly turned out to be like shooting a mosquito with a howitzer. Because honestly... what for? All the required services, Prometheus, exporters, metrics... And who will realistically look at this data? Pointless.

N8N Automations - huge capabilities, but I very quickly came to the conclusion that... I don't really have much to automate! Thanks to keeping things simple (more on that in the final conclusions), I simply have no need to use these types of tools. They weren't needed before, and I believe that instead of implementing them, it's worth considering what we did wrong that we suddenly need them - although, of course, I don't rule out scenarios where they genuinely help us with something.

Secondary self-hosted services like Linkwarden, Miniflux (rss) etc.: a matter similar to the above - after a moment of using them, I concluded that this is not for me. It simply makes no sense to overcomplicate your life when you don't use them. Bookmarks in the browser, a simple RSS, and everything you need is in its place. Without the entire baggage of configuring, maintaining, and securing.

A personal cloud like Nextcloud: in theory, this offers huge possibilities, but for me, just the installation itself was relatively problematic. Nextcloud is heavy, seems quite unpolished, and the same problem appears as with Google Photos - how am I supposed to convince others to use it? Just for myself, alone? Again, form over substance. It simply might not make sense.

Local AI: this can actually be very useful, provided we approach it smartly. The first tests involved OpenClaw and... never again. This service makes an incredible impression, but when, following a voice request sent on a messenger, it logged via SSH from one computer to another to check running processes, I decided that it is better to "pull its plug from the socket." However, I can imagine using AI and free models like Gemma as a sort of overseer for the homelab, our assistant. It could monitor things for us, let us know, or even independently fix detected problems. The thing is, it comes with great capabilities and great responsibility; we don't know if it will always work correctly.

Portainer for container management: I immediately approached the matter quite minimally, i.e., a repository with docker-compose stacks + sample envs. This in itself offers a backup and versioning for part of the infrastructure configuration. The whole thing was handled from the CLI. Good, but Portainer proved more convenient at times. A free upgrade to the Business version for 3 devices is possible, which allows you to control several environments and stacks on them from one central place. You can automate updating stacks after changes in the repository, etc. Very convenient, and I have no major complaints here.

NAS server as a backup for home devices + an extra USB drive plugged in additionally copying data to it + acting as a synchronization for OneDrive and Google Drive: after setting it up, all this worked correctly, but the issue boiled down to the fact that I realistically needed a max of 200 GB... and that can be handled via a DAS, even that same drive plugged into the main computer via USB, and performing backups manually or automatically (e.g., borg). So why configure a whole SMB/NFS, have a major dilemma about what to use at all, and why additionally synchronize it when it can be done much easier?

Home Assistant: a very popular application lately, full home automation, etc. I, however, rejected this idea very quickly due to the lack of compatible devices. Thermostatic radiator valves? After properly setting up the central heating system, this doesn't interest me at all. External window roller shutters? They require an additional central unit, and I have remote control access anyway. Air conditioning, washing machine, dryer? There are trivial-to-use apps on the phone. Only some sensors, e.g., smoke or water, might be useful here, but rather as something extra, and it can likely be done without a complicated system.

Monitoring, camera system: it crossed my mind slightly to somehow utilize the space available on the NAS, but I quickly dropped the idea. Why? I've lived where I live for two years now, and throughout that entire time... absolutely nothing has happened. Once in winter, out of distraction/oversight, I even left the garage door open all night, with bikes and some equipment inside. Of course, nothing disappeared. It's a quiet neighborhood, and as a last resort, there is already an efficiently working "neighborhood monitoring." My neighbors are fine, I live in friendship with them, so I decided that surrounding myself with cameras is pointless. Let's add that simply installing cameras isn't everything - you have to make sure that the chosen models aren't sending data to weird places, that everything always works, etc.

Conclusions

As a result of all this, I realized that I realistically have very little data that requires attention, backups, or a special approach. And most importantly, I definitely do not want to increase the volume. I don't do graphic design, I don't edit videos, and besides, I don't want to push too much into the "digital" world because it means I become more dependent on it. I'll give a simple example here: instead of having 100 similar photos in the cloud, I'd rather pick 2-3, print them, and have them in a physical album - a cloud backup is fine too, but there is no point in keeping 100 photos there. To me, for many people, the excuse "I have 10 years of photos, so I need a NAS for it" is just that - how many of those photos do you really need? How often do you go back to them? Where does the reluctance to finally organize them come from? Perhaps from simplicity? 100,000 photos or as many movies stored on a drive is still just one physical drive, and not like in the past, a whole room cluttered with discs or cassettes. Simplicity, but also something that truly consumes us. This is a somewhat philosophical issue, but I genuinely believe that we live in times of a huge overabundance of available content and we must filter it so as not to go crazy.

In my opinion, self-hosting is definitely not for everyone. I get the impression that it has become a fad in the last few years. Partly as a protest against big tech and their actions - which I strongly support, but that's a topic for a separate post - and partly, I think, because access to information is even easier, for instance through AI. The problem is that if we don't approach this really solidly, we can have problems. The basic one is data loss: there is no backup, and ideally a 3-2-1 backup system with proper verification, so in the event of a failure, our data is lost. But what if, due to our own mistakes, we lose access to important services because, for example, we self-host our password manager? What if all the emails we send go to spam because we decided that a personal server is a "simple matter"? And what if, due to weak security or a hole in the software being used, someone breaks into our network, takes over all or part of it, and then uses it, for example, for a crime? In such a situation, we also bear responsibility, because we are not just the user, but also the owner of this infrastructure. I have the feeling that many people forget this and think it's easy. Well, it's not. The prices of various cloud services often look the way they do because it all costs money. For a bit of contrast: I've had a Gmail account for... over 20 years? I set it up back when invitations were required. Throughout all this time, I have never lost an email. I'm not claiming here that no strange cases happen, because they do take place (outages, blocked Google accounts, etc.), but you have to keep in mind that behind such a massive company stands proper infrastructure, redundant backups, etc.

Here we come to another issue: cost-effectiveness. A homelab or self-hosting definitely does not pay off, unless we have a lot of data (several TB) and some good way to create backups, e.g., sharing space with friends, setting up a second server with family, etc. Despite everything, even then it can be very deceptive. Why? Initial hardware costs are one thing (and right now drives or RAM are really expensive), you also have to add depreciation, because everything wears out over time. You have to add the costs of electrical energy: the more active equipment we have, the higher they are (and I write this having a sizable PV installation!), and the more heat as well, which in some situations can become a problem and require cooling, air conditioning. On top of that come the costs of backups - the variants mentioned earlier are hardware, and in other situations, for example, paying for services like Amazon Glacier, Backblaze B2, or Hetzner... When we sum it all up, it quickly turns out that the costs can be much higher than in ready-to-use services.

But that's not all. We pay not only with money, but also with our own time. We will never get that back. Of course, we can say that it's our hobby, and I fully agree with that. Nevertheless, when I asked on Reddit how people manage a homelab/self-hosting, work, their own home, and above all relationships, family, most of the answers were... well, from people who either live single or don't have children. I used to be like that too; nowadays, I cannot imagine such a life anymore. Despite my passion, despite my love for IT, I must say openly: it has never given me as much happiness and joy in life as my child gives me every day. Now I rank going for a walk together higher than tinkering with another configuration. I am increasingly shifting things from the "done for myself, experimenting" category to "it just needs to work and not waste my time." A boring approach? Not at all, it's simply adulthood... and responsibility. Instead of a homelab, I prefer to develop a little human.

Last but not least: I must watch myself with the ideas that pop into my head. During yet another configuration of a various service, I wondered - could I have used this time differently, better? Of course, the answer isn't simple. Back then it was interesting, back then it was a challenge, so I really wanted to do it. Over time, however, it turns into tedious maintenance. And there is nothing interesting left in that for a person like me, a person who likes challenges, novelty, and opportunities for development.