Why the Home Cloud is two devices instead of one Draft
The obvious design for the Home Cloud is one box. One box is cheaper to build, cheaper to ship, and easier to explain. Every instinct I have about keeping things simple says one box.
It’s going to be two, and the reason has nothing to do with performance.
Parental controls have to survive the child
The whole promise of Curfew is that bedtime holds. Not that the app nags, not that it sends a parent a report — the internet stops. That only works if the rules live somewhere the child can’t reach, which means at the point traffic leaves the home.
A separate router sitting between your ISP’s modem and everything else has that property for free. A phone can be factory reset, a VPN can be installed, DNS settings can be changed, and none of it matters, because the rules are applied on the wire rather than on the phone.
Put that same job on the general-purpose server that also runs the photo library and you get a different machine: one doing enough work, running enough software, and restarting often enough that “the internet stopped for the whole home” becomes something that happens on a Tuesday afternoon for no reason a parent can explain.
The two jobs want opposite things
This is the part I’m most confident about, because I’ve lived on the wrong side of it.
The router has to be boring. It should run for years, do very little, update rarely, and never surprise anyone. It’s infrastructure, and infrastructure that’s interesting is infrastructure that’s broken.
The server is the opposite. It gets new apps, new updates, more storage over time. It’s the part that should change every month for years.
Those are contradictory requirements, and putting them on one board means one of them loses. In my experience it’s always the boring one that loses, because the interesting one is where all the reasons to touch it are.
There’s a plainer version of the same argument: if the server dies, the home should keep working. Wi-Fi, filtering and bedtimes carry on while a replacement is in the post. That’s only true if they were never the same machine.
What it costs
Two devices is worse in every way except the ones that matter. It costs more to make. There are two things to plug in instead of one, two power adapters, and a setup flow that has to explain a chain — modem, router, server — rather than a single box on a shelf.
I’ve gone back and forth on it. The deciding argument was the one I’d want if I were buying it rather than building it: I don’t want the thing enforcing my kids’ bedtime to be the same thing I’m installing new software on.
To be clear about what this is
This is a design decision, not a progress report. The first prototype server is on the bench now, with the Guardian router next, and the Home Cloud page tracks exactly where things stand.
What I can say is that the argument above isn’t theoretical. I’ve been running the pieces at home for about twenty years, and the separation is one of the few things I’d keep exactly as it is.