It’s a homelab. I start from scratch about every 2 years just because I can.

The only consistent element is the data on the system… the operating system, applications and frameworks often get a workout.

Having said all that, my current setup is working exceedingly well for my use case and I don’t know that I’m going to change it en-masse any time soon. Individual apps will change though I’m sure.

It depends greatly what you’re doing with your computer. I’ve been an Ubuntu primary user for at least a decade now, but there are still one or two things I can’t do with Linux… at least not as easily. Specifically in my case I have a PC dedicated to my synthesiser/MIDI setup simply because drivers, tools and plugins are readily available and supported on Windows while in Linux the support is good but has issues. A prime example is that my main synths are Roland synths, and Roland has a bad habit of not supporting standards fully for audio or MIDI over USB. I love my Roland synths but it made it too difficult to go Linux primary. That and some of the best tools are Windows-based like Ableton. However, that’s not to say you can’t do all of this in Linux; Renoise is a fantastic DAW as well that runs natively in Windows or Linux and there are plenty of great audio editing tools… it’s just interfacing with external gear is sometimes problematic and again some of my synths are plugins (VST’s) that either don’t work or aren’t well supported in environments other than Windows and Mac.

If your primary use case is web browsing and maybe gaming (Steam works great with their Proton runtimes for most games) then you can easily switch to Linux and never miss a beat. Firefox and Chromium (not Chrome) are great and well supported, and while Microsoft Office isn’t directly supported you can run the web Office365 just fine or install LibreOffice and still mostly be able to edit documents and files. For photo editing and the like there’s GIMP and if you’re a photographer there are amazing tools like DarkTable that are absolutely brilliant and in some cases are so good I would run them on Windows as well.

Ubuntu is probably one of the most mature and well-supported distros out there. Mint is also good and you won’t have any trouble with it, but when you do have issues with it the amount of support you can get for Ubuntu is among the best. There are things that Canonical do that annoy the Linux purists (systemd, Wayland as prime examples) for the average user these issues are pretty much moot. Even I who first installed Slackware Linux in 1993 and liked it don’t really care too much because the operating system gets out of my way and just works. I have one tool that requires me to run Xorg instead of Wayland (the GUI framwork… it’s complicated) but that’s literally the only issue I’ve had.

HTH

This has been a great discussion here… let me add a few things from my perspective of 30-odd years in the IT space;

• I like to use stuff that’s fit for purpose. Windows 10, Windows 11 and such are desktop operating systems that are fit for their purpose and are very good at it. But they’re less optimal for server-type workloads. Microsoft themselves provide a different operating system for that purpose but it has a different cost model that is a lot higher.
• Access to the GUI is necessary to run Windows. NAS devices and such have the ability to run “headless”; that is no keyboard, monitor or mouse. NAS devices also have a “network first” mentality where everything must be accessible on the network even in the event of a system failure. Recovery cannot require a monitor if you can’t plug one in! Windows (even server) requires physical access.
• Server-focused platforms like NAS provide a lot of capabilities that Windows does not because of the nature of their platforms. For example Synology allows growing your storage easily while Windows requires a lot more technical knowledge to accomplish that.
• Going back to fit-for-purpose; NAS devices provide security that isn’t necessarily there with Windows. Windows has a lot of “moving parts”… in addition to the operating system there are a bunch of ancillary libraries, tools and software that may or may not be used when using Windows as a server. All of these additional tools and libraries provide another potential vector for security breaches especially if not individually maintained thus increasing the maintenance requirements of the system. NAS devices give you the basics of what they need to operate and no more… well that’s until you start adding service packages to a Synology. But even then they will all be managed through the stock package manager and thus updated and maintained, and will still only be as much as you need to get the job done.

As far as my most recent experience with desktop Windows that I find irritating, there are a couple of reasons I still wouldn’t use it as a server platform ;

• Microsoft has a tendency to randomly update your settings, overriding your own settings with what they think are better. A good example that hit me recently is that some recent update overrode my power management settings on a PC I have set up as a headless desktop I then connect to using NX. I had it set to never sleep… suddenly it started sleeping. I had to reset it in order to get back to where I wanted it. This is not the first time this has happened, and I’ve had other issues along these lines. 24x7 isn’t possible when your PC goes to sleep…
• Windows lacks a really solid local filesystem. NTFS is OK and is pretty performant but it lacks a lot of the more advanced features of filesystems from NAS vendors or *NIX systems; ZFS and others have checksumming and scrubbing, most NAS vendors allow scheduled data integrity checks and the like… things like that.
• Software RAID in Windows is acceptable, but is not great. It’s hard to understand when things aren’t working properly and thus plan to replace failed hardware.

Hope that helps :)

Personally I DO self-host… and I have very few problems. I get blacklisted occasionally but it’s not been a huge concern and is usually only the low-priority blacklists… I did have to go through jumping through hoops early on to get my IP accepted but I haven’t had problems in years.

For my mail server these days I use Docker Mailserver. It’s really complete as a server (no frontend though) for setting up a really good IMAP/SMTP server. I have a full docker swarm cluster running here that keeps it VERY reliable. For a frontend on my desktop I use Evolution or Thunderbird (I’m a Linux user).

For a web frontend I have a few I have played with. My current “primary driver” is Snappymail acting as a plugin to my NextCloud instance. However I’ve had good experiences using E-Groupware which is VERY feature complete as an Outlook alternative.

Hope that helps!

It really amazed me how quickly my batch 2 arrived. I had to make an “emergency run” to my local MicroCenter (I’m fortunate enough to have one about 10 miles from my house… and maybe unfortunate for my bank balance…) for an SSD. Thankfully they had exactly what I wanted in stock and I was ready to go to the races LOL

I posted this in another comment but wanted to expand on it in a direct reply;

It’s worth noting that a big problem with ARM is it’s not one unified architecture. x86-64 has both the problem and advantage that there are more than one manufacturer of CPU’s, but all of them are boxed into being compatible with Intel. This isn’t true with ARM; while the CPU architecture itself is nominally the same, individual vendors are free to implement that standard in myriad ways with myriad support chips that they can pick and choose or develop themselves. Something written for NVidia ARM won’t run on Qualcomm and vice versa, and neither will run on Mac. Unless stricter standards are wrapped around the architecture we’ll never see a true “generic ARM architecture” system. But that’ll never happen beacause it’s a violation of the terms of the agreements that the fabs signed with ARM.

If a single vendor came along with an architecture design for ARM CPU that everyone liked and that vendor allowed other manufacturers to create compatible designs then we could see that standard appear. Maybe Framework could be that vendor… but right now they’re FAR too small with FAR to small pockets to develop an architecture of this magnitude. It’d have to be adaptable to laptops, desktops, datacenter, embedded… you name it. This would require a company with deep pockets and a LOT of engineers on staff. NVidia has both of those but is NEVER going to publish a standard so that Qualcomm can create a competing CPU. Apple’s not going to open their architecture up either for the exact same reason.