If a drive goes down and is being rebuilt, all data being retrieved from that RAID array has to be constructed using RAID parity. (In RAID 4 or 5, the data is striped across all drives in the array, so a "bit" if every data record is on every drive.)
But that computation is very easy, using parity. When using odd parity, the data on the "good drives" is read and if there are an even number of 1's, then the missing bit is a 1. If there are an odd number of 1's, then the missing bit is a 0.
There is virtually no performance degradation from this computation. Performance degredation is simply from the system using I/O and CPU resources to reconstruct the failed drive, which is controlled by the sys admin.
If he is using "striping" then the same applies to his new machine.
NTAP filers are very fast with reads, too. There is no inherent advantage to mirroring versus RAID in reads. NTAP's read caching algorithm is very good and its performance depends mostly on how much RAM is configured in the filer.
BTW, this performance during rebuild issue is almost moot, as the occurance of a drive failure is extremely rare, especially since the filer reports soft failures long before a hard failure brings down a drive.
There is more to your friend's story than we are hearing. I can't imagine what they needed that the filer did not provide and I can imagine lots of things they need that only the filer provides. |
