I've had customers running SATA raid arrays for years. I've always noticed a slightly higher rate of failure with SATA drives, then SCSI drives in a standard setup. Recently, I've had a few customers where their original SATA configurations start losing disks at a very high rate. When SATA raid first came about, there wasn't much information regarding what SATA disks to use in a raid configuration. Now, everywhere you look manufacturers are offering enterprise level disks.
There is a very important difference between these disks and standard desktop drives. The error recovery feature. You see, standard desktop disks use error recovery, to the extent that a drive can be unavailable to the OS for many seconds. This is ok in a standard single disk setup, as the OS will just wait for the disk to become available. But in a RAID configuration, the controller thinks that a disk that is not responsive is failed, and will mark the drive as dead. In a system in a higher load the problem becomes more prevalent.
For any SATA raid controller, be sure to purchase disks that have the error recovery feature disabled. For Western Digital, that would mean disks employing their TLER technology, which currently is their RE line of drives. Who would guess it; RE stands for RAID Edition.
Better yet, find documentation, or ask the manufacturer of your controller what drives, and firmware levels have been tested for compatibility.
0 Trackbacks