Here is more recent paper from Dr Steve Hetzler, an IBM research scientist
The proponents of ZFS forget to mention limitations such as not being able to dynamically expand/reduce pools, the almost requirement to have a fast ssd as a cache drive, complexity in managing it etc. It is no doubt an industrial strength file system, and it is probably out of place for many home uses unless the user is advanced (which many here are).
I also would base the decision not on the size of data (2TB or 200TB is immaterial) but on how static it is. If its mostly media which is write once, then a snapshot solution is actually better as it doesn't stripe data and is much simpler in theory and in practice, as well as being infinitely expandable. For me, simplicity is paramount - both in the technical implementation as well as the management interface.
If there is ever an implementation of ZFS that is easy for the end user (by which I mean no cmd line needed, ever) and which is more flexible, then it might be an alternative. Storage Spaces is dead in the water and btrfs will only be in Linux which rules it out. Meanwhile, I can take 10 2TB drives and store their parity on 2 more drives this giving me reasonable peace of mind as well as disaster recovery - I can simply pull out a drive and use it, even if there's bit rot only a few affected files are damaged. This is why I hate striping solutions.