You're viewing a single thread.
The overhead of duplicated data across the network. Not reposts on different instances, but the software itself on those different instances needing to cache/store this one single post for their users locally
Don’t large services have many duplicates/caches spread across the globe to balance load and reduce latency? Couldn’t this be seen as a positive? It could also be seen as a redundancy layer.
Yes. It's very common to cache content closer to the user, otherwise the site would be slow. Some services like Netflix and Facebook even provide custom caching servers to internet providers to install in their data centers. These are called Netflix Open Connect and Facebook Network Appliance respectively. They significantly reduces costs for the ISP, as Netflix and Facebook are generally two of the heaviest users of bandwidth on an ISP's network, and traffic entirely within their own network is effectively "free" for them.
This is a good part of federation IMO - if users join an instance physically close to them, their experience is going to be nice and fast, since everything is cached on their instance. It's also pretty easy to spin up a new Lemmy instance in your country if one doesn't exist yet.
it's absolutely just a redundancy layer, in fact this is one of the main benefits i see with federation.