tend to compress and dedupe very well

do you have any pointer?

by my term cheep it is comparing to solution nvidia DDN node

and what do you mean by the term “Once you delve deep into these questions”? because this assume i have some general knowledge about the storage stack that can scale to petabyte level storage. because i bet it safe to assume that truenas scale won’t scale to a petabyte or more

ML dataset,
i have no plan for software for now

yes it dose need to be an enterprise system it onsite

this must be self hosted

for the buget it provided by my uni