The software architect of lemmy is unfortunately doomed. The very concept of how it works means exponential storage and bandwidth needs as it grows in sublemmits and instances.
A better design would have been instances being the sublemmits themselves, and leaving it up to the clients to subscribe and aggregate them into a feed. This way scaling is a lot more horizontal, and communities that get too big can scale up individually or purge old data without affecting the rest of the system.
If a user of an instance subscribes to content from another instance, their home instance is pulling, storing and sharing that content. With more and more instances, more time will be spent on sharing that content.
If the design itself is bad, then something will eventually spring up that will replace it. That’s the beauty of nascent platforms; they haven’t completely cornered the market.
The software architect of lemmy is unfortunately doomed. The very concept of how it works means exponential storage and bandwidth needs as it grows in sublemmits and instances. A better design would have been instances being the sublemmits themselves, and leaving it up to the clients to subscribe and aggregate them into a feed. This way scaling is a lot more horizontal, and communities that get too big can scale up individually or purge old data without affecting the rest of the system.
I assume this is a larger theme across the Fediverse?
Could you expand on what causes the massive bandwidth needs? I’m have a vague idea but I’d be very curious to know.
If a user of an instance subscribes to content from another instance, their home instance is pulling, storing and sharing that content. With more and more instances, more time will be spent on sharing that content.
If the design itself is bad, then something will eventually spring up that will replace it. That’s the beauty of nascent platforms; they haven’t completely cornered the market.