I prefer good faith discussions please. I love the Fediverse and love what it can be long term. The problem is that parts of the culture want nothing to do with financial aspect. Many are opposed to ads, memberships, sponsorships etc The “small instances” response does nothing to positively contribute to the conversation. There are already massive instances and not everyone wants to self host. People are concerned with larger companies coming to the Fedi but these beliefs will drive everyday users to those instances. People don’t like feeling disposable and when you hamstring admins who then ultimately shut down their instances that’s exactly how people end up feeling. There has to be an ethical way of going about this. So many people were too hard just to be told “too bad” “small instances” I don’t want to end up with a Fediverse ran by corporations because they can provide stability.
As a small instance owner, I can say its not sustainable. I’m paying approximately $40 monthly out of my pocket but with it, I have a non-defederated, long-term instance. So I don’t mind much, just enjoying Lemmy.
I think the real problem is not financial, but technical problems. For example, when a post is shared in an instance, that instance sends that activity to approximately 1000 other active instances. As a result, as the number of instances increases, the load on the network also increases. Ironically I think it should be the other way around :) I'm not sure how it will scale in the future.
In addition, since each instance keeps all the data in its own database, the database size of the instance with 1m users and the instance with 1k users is the same. In my opinion, this is what is really unsustainable.
Though I was technically naive about how the fediverse could work, I was generally curious … but when I found out it’s distributed data synchronisation it was one of those moments in tech when you realise something isn’t that fancy and is done essentially the way most people would do it if they had to design it.
My presumption was that there was some robust but efficient network of servers that aided cache and data retrieval.
As you say, data synch seems to put a decent load in all servers which grows as the network does. Seems like a problem that’s been kicked down the road. As you say.
Well… KISS is a good concept for a reason. There's on paper no reason - at least before running into actual issues - to find something fancier than just replicating updates across known other network members.
Oh I know. But by 2022 the Fedi had been running for years so I’d figured something fancy had at least become necessary by that point. How much of a problem it is I don’t know but I’ve definitely heard that the load of pushing out all the sync has been significant for some servers, lemmy instances included during the migration. And then there’s the storage costs, where that’s probably a much smaller problem but might come up at some point not far down the line??
It'd be neat if there was some form of peer-to-peer activity-push to resolve this; basically offload your pushes to other instances, and in return they can offer some of theirs to yours. I think that gets quite difficult though, especially as large lists of federated/whitelisted/defederated instances come into play.
Hehe, I wonder if some technology like Tor could help here, distributing the work across many servers instead of each one having to do all the work entirely by itself? One day, if someone wants to build it, that may come, but it will not be me, nor today it seems:-P.