Yes, that should work.
Since I'm a developer and not an enduser, I'm not dealing with such an amount of data. My experience is that whenever users have such big databases they also do have a lot of junk in it, for example files that are not referenced anymore in the database file but do still persist on the hard drive. After time they are wondering why random studies won't be stored locally when querying a pacs (because these files and folders exist already and thus they will not be registered in the database file).
Best practice is to make use of the overflow management and regenerate/ delete the database from time to time.