That however would be a different problem. A horizontally scaled instance would be able to cope with more users, but if it shuts down for monetary, personal, or whatever reason, it’s still down.

Protecting a community from this is what the decentralized part is for. That is already in place.

(Although there is a middle ground where you could design the system in a way that one instance is mirrored and load-balanced across different hosters. That would actually also be quite interesting to have. But that’s another layer of complexity on top.)

I think lemmy will be bitten in the ass by not having considered clustering/horizontal scaling from the start. Federation alone as a scaling mechanism is only feasible for “nerds”. But if the network wants to grow, we will need a few scale-able large hosted instances. And if their only choice is to scale vertically, there will be a hard limit (unless we put a good old Mainframe somewhere ^^).

Another downside of this design is: you can’t run it with high availability. If there’s only one process per instance, updating it will mean the whole instance is down. Sure, if all goes well this downtime is under a second. But if it doesn’t go well or if a migration is needed, this might quickly become hours.