no need to have the OS or BDR on it. not recommended unless doing a raid 5 for OS.
if i had that controller and the drives i would create 2 x 8 drive raid 6 with hot swap spares
use 1 raid array as your media etc drive
1 raid array as Export
and i would still back up externally :-)
I would suggest making a work array and a archive array or mirror a raid 5 or 10 to the 2nd array.
Work array - 10 Drive raid 10
Archive array - 10 Drive raid 10. Raid 6 if you use the Enterprise drives.
1 or 2 drives for hot spare.
Work. - 12 drive raid 10
Archive - Mirrored 12 drive raid 10 array for backup.
Let me add one thing I would do before setting up the final array's. Buy a copy to HD Tune Pro and initially test your disk drives with RAID 0 and see that you have good read and write performance. I have found that using a large quantity of end user drives that sometimes you might get a drive that just will not perform properly in an large array although it works perfectly as an individual drive. The larger the array the more likely you will have a disk drive that will not perform properly, my results showed a significant drop in write performance.. Hunting for the drive that drags down your performance is a trial and error process. I have also found that the first 8 ports of the Areca controller are native ports and provide the best performance. The other higher number ports will not give as good performance as the first 8 because they are achieved by SAS/SATA switches off the first eight.
Here are examples that I have seen with 8 each 1 TB drives with fairly good write performance and with bad write performance.
Bill Gehrke wrote:
I have also found that the first 8 ports of the Areca controller are native ports and provide the best performance. The other higher number ports will not give as good performance as the first 8 because they are achieved by SAS/SATA switches off the first eight.
Thank you for that! I think you have just answered something that has been confusing me on my 12-port Areca.
How did you determine this, or did you read about it somewhere?
Loads and loads of testing. I initially tried 16 disks 1 TB drives and got lousy results. I found two or three drives that affect performance. I pared it down to 8 drives that got good results so I could check the performance versus low/high ports, after I got and tested a "good" set of disk on ports 1-8, I moved them to 9-16 on my Areca ARC-1880ix-16 and found worse results just by changing the two connectors to the higher ports. Now you can see why I suggest quick initial testing with RAID 0 which has no initialization time. I still have to do a lot more testing on different RAID configurations to see what the optimum setup is under real RAID configurations. Of course I am simultaneously running PPBM5 tests. And another experiment I plan on is multiple controllers.