If they were batches retrieved from a subledger, you might not have to restore the whole dataset. Is that the case? If so, I'll let you know how.