Move to a different instance to spread the load. You can see all the same content from any other instance, but the experience will be a lot better with less lag.
Lemmy is not the same as lemmy.world.
The problem with lemmy.ml and lemmy.world is that they are just too popular and instances don’t scale well.
I don’t understand how a million tiny instances is supposed to scale better than a few big instances.
Caching all the data from another instance is overhead. If you’re not serving that to enough people, your instance is going to create more traffic than it reduces.
1-10 person instances can’t possibly help. Maybe 10,000 users on an instance is valuable for scaling.
I would indeed say 1000-5000 users instance should be the soft spot.
Having a look at https://lemmy.fediverse.observer/list and filter by 1m (monthly active users) shows that 27k are on Lemmy.world, while Lemmy.ml is second with only 3,8k.
A healthier solution would be to have all the small instances (imagine the 25 biggest, so up to Lemmy.zip) to gain users, so that LW would be less critical
Their ALL feeds will only have their individual servers unless their users go to the other instances and subscribe to communities. And only the communities they subscribe to will be fetched. But they’ll be fetched for all users on your instance with only one sub.
Each instance only downloads relieve content to its users. An instance with a popular community will have to handle all the posts made to that community, which will still be much smaller than all of them. While the overall load might be higher, the load on any given instance will be lower.
Images, by far the biggest bandwidth user, are directly transfered from posting instance to client and are not federated. Having more instances will spread out that load very effectively.
Helpful trick, if you go to [any instance url]/instances (for example https://discuss.tchncs.de/instances ) it will show all the instances it is federated and defederated with. (Use your browser’s find in page feature to scroll down to the “blocked instances” section)
Move to a different instance to spread the load. You can see all the same content from any other instance, but the experience will be a lot better with less lag.
Lemmy is not the same as lemmy.world.
The problem with lemmy.ml and lemmy.world is that they are just too popular and instances don’t scale well.
I don’t understand how a million tiny instances is supposed to scale better than a few big instances.
Caching all the data from another instance is overhead. If you’re not serving that to enough people, your instance is going to create more traffic than it reduces.
1-10 person instances can’t possibly help. Maybe 10,000 users on an instance is valuable for scaling.
I would indeed say 1000-5000 users instance should be the soft spot. Having a look at https://lemmy.fediverse.observer/list and filter by 1m (monthly active users) shows that 27k are on Lemmy.world, while Lemmy.ml is second with only 3,8k.
A healthier solution would be to have all the small instances (imagine the 25 biggest, so up to Lemmy.zip) to gain users, so that LW would be less critical
Is there any functional difference between 4 people funding and running 1 instances vs 4 people funding and running 4 instances?
Scale that to a few millions on 1 instances and you get Reddit, so there is at least a difference in terms of decision making
Their ALL feeds will only have their individual servers unless their users go to the other instances and subscribe to communities. And only the communities they subscribe to will be fetched. But they’ll be fetched for all users on your instance with only one sub.
(Technically they’re pushed and not fetched.)
Each instance only downloads relieve content to its users. An instance with a popular community will have to handle all the posts made to that community, which will still be much smaller than all of them. While the overall load might be higher, the load on any given instance will be lower.
Images, by far the biggest bandwidth user, are directly transfered from posting instance to client and are not federated. Having more instances will spread out that load very effectively.
Nah. I moved from ml to world to .ca, and .ca is the best. I didn’t realize how much content ml and world defederated from. I love it on .ca
Helpful trick, if you go to [any instance url]/instances (for example https://discuss.tchncs.de/instances ) it will show all the instances it is federated and defederated with. (Use your browser’s find in page feature to scroll down to the “blocked instances” section)