As someone who is currently implementing a next-gen distributed file system, I can highlight one aspect: you have a lot of concurrency and asynchronous processi

As someone who is currently implementing a next-gen distributed file system, I can highlight one aspect: you have a lot of concurrency and asynchronous processing. Thus you need at least reference counting.