With such large drive sizes, if any of the data is important to you, I would say Follow 2^n+1 rule. That is n parity drives per 2^n+1 drives or thereabouts. Try to keep the array size less.
If you plan to go with 16TB drives, I would say you'll need at least 16 drives total, and 2 vdevs of 8 drive RAIDZ2 each.
So you can afford to lose 2 drives per vdev.
Remember, RAIDZ does not stripe the data (unless you put drives/vdevs in stripe) so speeds will be limited to that of single drive.
Hence, if let's say your 192TB storage space was 30% full,
And one drive on a vdev died. So that's 16TBx0.3 = 4.8TB of data to be rewritten the replacement drive (resilvering). With sustained write speeds of 250MB/s, that's 5 hours 20 mins continuous activity where all 8 drives in the vdev are at 100%. Have heard and seen drives fail during resilver, hence the 2nd parity will help.
Also, TrueNAS uses RAM as cache, so yeah ECC RAM does help but is not a requirement. I would suggest get some 16GB optanes from ebay for $7 a pop and use them as L3 ARC (Exactly what I have done). Optanes, even though they don't have high sustained read or write speeds, have some amazing endurance and I/O rating. So they are perfect for TrueNAS boot drives as well. You can use mirrored NVME volumes as storage for containers and volumes for the apps you are planning to run on the NAS.
