LSI RAID controllers can expand array during use/box up running a business, so robust can even reboot thru if needed!
Can set performance hit for the background 'reconfig' of the array.
All ways and always have EXTERNAL backup; and be ready to go olds-cool and have backup to image to new array.
.
Should always have redundancy, HD is 1st point of fail, especially as old world motor in an otherwise solid state box.
>> so RAID_0 is out of picture
.
Parity RAID levels 5,50 aren't favored for OS, as must calculate parity for every write to HD's;
OS array, especially Windows has constant writes for virtual paging of memory, registry etc.
RAID_6 , 60 with double parity calcs of different types bogs box even more than RAID_5 etc. handling OS
Parity calc is a real overhead, best if use only on controller with cache memory to help with the overhead of the parity calc.
You can expect the RAID_5 to move faster the more drives/ wider the stripe like a RAID_0......
Until about HD_8 or 9 added, then, the parity calc is so big, array moves slower.
ON old RAID_4, was the same as RAID_5, only parity calc# was all on last HD, which got burned out faster, and made bottleneck
Parity is real overhead.
.
Mirror RAID_1 or 10 more expensive to deploy, but best for OS array, and then cheaper to deploy RAID_5 etc. for data on back end if mostly READS not WRITES.
.
HD's do make a difference, SAS have better motors, 2x as much circuitry board, full duplex, faster,less vibrations(cabinet of sata HD's can make slower array by 1 HD vibes making other HD arm take longer to track to data!) last like 2-3x longer etc. etc.
If you put a SAS connector on sata HD, lasts the same lifespan, put the SAS($) firmware on it and it seeks better, and lasts about 2x as long!!!
SATA, cheaper, half duplex, no err checking. SATA originally okayed in RAID_5 of 3 x 50g drives etc., but we are so far beyond that, rebuilding a RAID_5 on sata is really risking array fail more and more. SATA are more 8x5 weekly HD's, SAS are the real 24x7's. SATA in such a use usually 1yr warranty, SAS hd's generally more!
.
Scrubbing thru array with Consistency Check monthly and before reconfig advised, to make sure array consistent/pristine. Array typically single fault tolerant, so want 'perfect' array , so if HD fails etc. it is the ONLY point of fail in single fault tolerant strategy.
Else, have 2 faults and array punctures, then replicates that in parity block(if array still functions at all), so now 3HD's have fault...
If puncture array and puncture is in backup and redeploy to brand new everything, would still have puncture in backup and can bring to new hardware like bringing a VIRUS back. The VIRTUAL bad blocks on HD seen as PHYSICAL bad blocks, and can even throw predictive fail flag; that can't be withdrawn, and no other warning when HD really has HARDWARE fail imminent.
.
