That's a reasonable plan. And I 100% agree that controlling your data is preferable to having it mined.
What frustrates me, and why I don't do it any more, is we don't have a manageable story for most things. In many cases there's a good recipe you can follow to stand up the basic service. But hardening, security patching, etc. aren't covered. You have to come up with your own solution to make sure it's up, etc. On top of that, projects come and go, and someone may unfortunately choose a project that's a dead end and won't be patched. (And big-budget cloud doesn't solve many of these issues, either).
My personal fear is that a lot of self-hosted stuff becomes like all the unpatched Wordpress sites, years ago, that were just vectors for hacks. It wasn't that the data was stolen, they were pretty much pwned to launch other attacks. There are just too many solutions out there for all the bits and pieces needed to keep stuff up and secure. And all those fiddly bits are hand-integrated (for the most part). I'd like to find something that provided me a full stack, with all the boxes checked. I would get monitoring and security patching around all the bits.
In the interim, I try to use products from companies that either don't primarily make money by advertising based on my data (even if the products are more expensive). (Note that advertising is what you do when you're out of real ideas - so it's inevitable that all companies head that way when MBAs with no imagination want a safe return). Or, I use products that are (as much as possible) open source. (There are still disturbing amounts of proprietary blobs in my Raspberry Pi homely servers, for example).
With all that said, I wish you luck! I've run my own infra in the past and it's fun.
The experience needs to be more like installing an app on a phone or a game on a console and with automated backups and easy restore. Self hosting will never come back if we are still expecting people to manage servers like it’s 1985.
I think part of the problem is that this is exactly the kind of stuff that is not fun. It’s that boring middle layer between the OS and apps. Compounding the problem is the fact that the people who know that layer well feel no personal need to fix its usability issues.
Programmers want to work on AI and distributed systems and games and other sexy things. Even people who work in the middle layer would rather work on sexy problems there like hyperscaling with Kubernetes and architecture as code.
Making regular old systems easy to use for boring not-hyperscale uses just isn’t sexy so open source devs don’t do it. The economic model for commercial stuff only incentivizes the leveraging of this problem for vendor lock in or to move people into SaaS where data can be mined and rent can be charged forever.
At this point the whole industry is herding everyone into SaaS walled gardens because that’s the only working economic model in software. I don’t see this changing without a movement similar to open source in its heyday in the 90s, but to steal fire from the nerds and bring it to the masses.
I’m not optimistic because nobody seems to care. It might take a whole cycle in which all freedom and privacy is completely lost. Experiences like Twitter just aren’t cutting it. People are stuck on either “woke Twitter was bad” or “pilled Elon bad” instead of realizing that the problem is intrinsic to walled gardens. All cloud spies on you and all social media is manipulating discourse for someone. No exceptions.
> The experience needs to be more like installing an app on a phone or a game on a console and with automated backups and easy restore. Self hosting will never come back if we are still expecting people to manage servers like it’s 1985.
It exists: https://yunohost.org/#/. Install the distro, and then it's all clickety-clicks for all your apps (which are softwares like nextcloud, cryptpad, bitwarden, bitter, etc..)
That is definietely the idea, but in practice all the app stores suffer from same problems - mainly that keeping the apps up to date is a lot of thankless work.
Yunohost is mainly full of community contributions, quite a few of which have been abandoned. Some are stuck on old versions, some use migration scripts which may or may not do things the correct Yunohost way, some use migration scripts with bugs which can lead to data loss. The front end is slick, but it's the wild west behind the scenes. There's not even a mechanism for regularly reviewing if apps have been abandoned - I've manually reported a couple.
Cloudron is probably better than most as they have a financial incentive, but then that targets their apps towards "professional" users.
FreeNAS kinds of sits in that area, but it suffers from some of the same community contribution problem. It’s definitely hit or miss. I’m thinking the solution has to be like 10 core things that are rock solid rather than dozens of things that range from excellent to hasn’t-worked-in-2-years.
What frustrates me, and why I don't do it any more, is we don't have a manageable story for most things. In many cases there's a good recipe you can follow to stand up the basic service. But hardening, security patching, etc. aren't covered. You have to come up with your own solution to make sure it's up, etc. On top of that, projects come and go, and someone may unfortunately choose a project that's a dead end and won't be patched. (And big-budget cloud doesn't solve many of these issues, either).
My personal fear is that a lot of self-hosted stuff becomes like all the unpatched Wordpress sites, years ago, that were just vectors for hacks. It wasn't that the data was stolen, they were pretty much pwned to launch other attacks. There are just too many solutions out there for all the bits and pieces needed to keep stuff up and secure. And all those fiddly bits are hand-integrated (for the most part). I'd like to find something that provided me a full stack, with all the boxes checked. I would get monitoring and security patching around all the bits.
In the interim, I try to use products from companies that either don't primarily make money by advertising based on my data (even if the products are more expensive). (Note that advertising is what you do when you're out of real ideas - so it's inevitable that all companies head that way when MBAs with no imagination want a safe return). Or, I use products that are (as much as possible) open source. (There are still disturbing amounts of proprietary blobs in my Raspberry Pi homely servers, for example).
With all that said, I wish you luck! I've run my own infra in the past and it's fun.