> My recommendation to most people putting services online would be: either do it for yourself only, or do it as a team with proper structure and processes. What sounds like an initiative to emancipate people could actually alienate them to you, and that is a huge responsibility.
Oof, good advice. I run a startup that helps folks self-host, but it really does split the audience in two. Folks technical enough to swallow the somewhat rough edges become huge fans and part of a fun community. Folks just on the other side of that split tend to have pretty frustrating experiences...
I dearly wish I had the capital to be able to spend another full-time year on making our product better, but self-hosting is a really tricky thing to build a company around - the audience by definition is looking to avoid paying for services!
I do still fully believe (and hope!) that one day, far from now, self-hosting reliably will be trivial, and our kids will all think we were a bit slow for relying on a few megacorporations hosted services.
I think the main problem is that ordinary people don't even see what problems self-hosting is supposed to address; and those that do, still need to dedicate significant time and effort to "tinkering", even when handed a huge chunk of the solution on a plate.
Another huge problem is that there's a home network between your product and the user's other devices; most home networks are utter crap, and often even tech-savvy people don't have a whole lot of control over it (I hate my ISP's modem with passion). This seriously limits your potential to provide an excellent UX; IMHO it's the UX that makes or breaks a product for "the rest of us".
I used to self-host a whole bunch of things on a VPS, including my blog, git repos, a DIY blogroll / RSS reader, etc. In the end I've decided it was not worth the effort; the blog was moved to Netlify, repos to Github, and the RSS kludge got swapped for NetNewsWire with iCloud sync. I was paying €5 for the VPS, yet now I'm paying Apple €20 to host my email, sync my photos, get access to the music catalogue, etc. I would definitely pay €20/mo for a box under my desk + an online service, provided it gives me similar value without much additional effort.
I think the problem that KubeSail/PiBox is aiming to solve might be both too broad (run any software you like!), and too narrow (if you're an enthusiast!) at the same time. I don't want to run Miniflux; I want to have my RSS feeds synced between devices. The software that pushes the bytes (and the hardware it runs on) should be invisible - unless I decide (out of my own free will / curiosity) to pop the cover open and start tinkering.
I don't think you can solve this by addressing shortcomings in a single piece of the stack. Both the layer below you (your average home network), and above you (the apps) have their own problems; some are like splinters (tiny but enough to ruin the experience), some are fundamental ("what is MySQL and why do I need to know"). I don't think it's a lost fight, but I would try to start with a vision for a more vertically integrated solution; maybe one step of that road is to eventually build your own WiFi AP/router (or even become an ISP), maybe to make a deal with Spotify (or even directly with EMI/WB/etc)... I don't think a task is too big if you can seriously challenge Apple/Amazon/Google at the end of the road.
I agree! Unfortunately, we pivoted to self-hosting right around the time we were running out of money, and around the time I had a child and thus, needed money. I'm really glad we pivoted to something we love and our users love, but it hardly pays the bills.
I've spoken with several people who are starting similar companies and who've reached out to me (happy to do that!) - my advice is similar to yours: keep it simple, keep it focused. KubeSail is a developer tool turned home-hosting tool, but if I could rebuild it, I'd make it incredibly simple to get Jellyfin and a torrent/VPN client installed and that's about it, and then execute insanely hard on making that as streamlined and foolproof as humanly possible.
I think if you could sell an as-easy-a-chromecast box that could do jellyfin, had a nice ui for uploading local media, and had an easy guide or built in VPN/torrent client, you’d be to build a great business.
Of course - you can’t exactly vendor torrent stuff - and I’d never suggest anyone to pirate anything. But certainly the sky is the limit, and that’s just media. Other tools like Monica CRM, Tandoor Recipes, Mastodon, etc are their own markets too!
We’re too far in the technical side to be mass appeal, and our UI/UX is far from “mom-friendly”. Still - I’m optimistic a better entrepreneur than myself will conquer this one day.
There's already a huge market for SOHO network gear, NAS / media appliances, etc. and all of that needs to be usable by non-experts. We need to ask the other question: what motivates people to self-host?
It's a modem+router+switch+AP that technically does everything you need, but does all of it badly - really just getting in my way. E.g. it obviously has a builtin DHCP server, but it won't let me set custom DNS. I want to use custom DNS, to block at least some of the ads/tracking on all of the devices on my network; so I have to disable that DHCP and use my own. But the modem resets that setting back to enabled every reboot! (Took me a while first time around to notice there's two DHCP servers on the network, argh). So I've disabled the internal AP, brought my own, and I'm connecting the rest of the network through a managed switch that blocks DHCP to the router.
So I've got one device that tries to do the job of four... but instead I need three devices to do the job of one. I try not to think about it.
Yeah I just have hardware I like running openwrt behind the modem doing all three jobs, with Tailscale Funnel doing the job of port forwarding so I basically never need to interact with the provided modem's interface. Because you're right, it sucks.
The problem is the builtin cable modem. I don't have the parameters/credentials, I have next to zero experience wrangling DOCSIS, and frankly I just don't want to bother. Switching to fiber soon enough anyway.
This is typical of ISPs in France: you get a crappy device that does router + switch + wifi but none of them are done right.
There is a whole community around replacing these devices with other devices (owned by the users) but ISPs make it difficult despite a European law that forces them to allow this.
I used to have an Orange box that was hanging on a regular basis when there was too much traffic. Had to reboot it nightly.
I know have a high-end box from another provider that is OK, but I still do my own DHCP.
Yes. It is really not bad. I used to be in the "replace your box" camp (I replaced the shitty Orange box) but with the Delta I use it as expected. Not to mention that it can run VMs so I put Pihole there too (in addition to my home server).
It’s been a long time since I’ve used my ISP’s supplied modem/router. The only hardware from my ISP that I’m using now is their GPON SFP module, which is installed in the NIC of my router machine.
About my business? Sure! It's at https://kubesail.com and we sell our hardware at https://pibox.io (the software works with almost anything that can run Linux tho!) :)
Our best feature is that the website will detect if you're on the same network as your machine and if so, offer "local" links instead of remotely proxied ones. That way non-technical users dont need anything fancy or to be aware of how NAT traversal works. On top of that, the "local" urls still get valid HTTPS certs for free, so non-technical users dont get any scary browser warnings.
We started out as a way to make self-hosting easier for corporations, and were doing consulting work, but the users who joined our community were mostly home-hosters, so we leaned into that! Jellyfin is now our most popular app.
Given the market that you're after, why sell it as a SaaS? The people that want new subscription services, and the people that want to self-host feels like an empty set. Why not do the more traditional model of selling version 1 of the software for $x, and then when version 2 comes out, sell that for $y, and people with version 1 can pay $z to upgrade, where z < y.
The math could work out to be the same, but the psychology of marketing is everything. If I, as a hard-core-self-hoster, pay $60 for a version 1 of software that I can use forever, and version 2 comes out a year later, and I pay $60 for that; I'm much happier to do that, compared to having to pay $5/month for yet another subscription service, even though that's exactly the same amount of money. I already have so many subscription services! I don't want to pay for another one!
That’s effectively the tactic with the hardware - however - home hosting does need a SaaS imo! Between remote access (proxying ala cloudflare tunnels), backups, VPN, there are nontrivial ongoing services that need to be in place for happy home hosting.
That said - you’re not wrong at all and that’s why our service is totally optional and our free tier is quite generous.
The saas for sure made more sense when our target customers were companies.
The UX for a non-technical user to setup a port forwards is a total non starter, but self-hosted remote access and VPN are well covered by Tailscale + Tailscale Funnel these days. Backups for data is an actual service that people pay (and could get paid) for.
If you could provide similar functionality / workloads as the old MS Small Business Server isn't there a MSP / reseller market opportunity? Self hosting for businesses that are < 25 users. They still tend to be serviced by small MSP. They've gone cloud because that's where their vendors went. Many hate it.
You'd need
* Directory/Authenticaion
* Email
* Shared Folders
* Wiki / Web something
If there were options for adding IM, PBX, CRM it could be a compelling offering for resellers.
> 5-bay and desktop HDD compatible models are under development and will be coming soon.
The box does look pretty. Any plans for dual/multiple ethernet versions?
At a quick glance the Pi compute module doesn't have any so you must have added the lone one yourselves?
And of course the geek in me would like to know the network chips and how they're connected to the compute module (although I guess usb is the only choice).
For me self hosting is about taking responsibility and having the option of identifying and eliminating faults and threat surfaces myself. This is very much extra that will consume time, energy, and attention. With all this my expectation is self hosting ultimately amounts to premium service which is likely to cost more starting with the hardware and management of it. Seems like this is not a big sector, but a potentially profitable one with some very committed clients.
Would you happen to know why your customers choose to self host? There's a myriad of potential reasons, and I'm curious which ones are the primary ones.
Oof, good advice. I run a startup that helps folks self-host, but it really does split the audience in two. Folks technical enough to swallow the somewhat rough edges become huge fans and part of a fun community. Folks just on the other side of that split tend to have pretty frustrating experiences...
I dearly wish I had the capital to be able to spend another full-time year on making our product better, but self-hosting is a really tricky thing to build a company around - the audience by definition is looking to avoid paying for services!
I do still fully believe (and hope!) that one day, far from now, self-hosting reliably will be trivial, and our kids will all think we were a bit slow for relying on a few megacorporations hosted services.