To fill out the DS-412+, I picked up four 2TB Western Digital RE4-GP enterprise-class SATA disks. The choice to go with the more reliable enterprise-class drives was an easy one; I'd experienced probably one disk failure every three or four months with my DroboFS, which was populated with 2TB Western Digital Green disks.

ERC/TLER is vitally important for any spinning disk drive used in a RAID set, because it controls the amount of time a hard disk drive controller waits when it encounters a read or write error before it reports the error to the host system and begins attempting to recover from it. Without ERC, a disk drive in a RAID set which runs into a bad sector on read could find itself dropped from the RAID set entirely by the RAID controller, which could in turn trigger a full RAID re-build to a hot spare (with the performance impact that incurs) or could simply leave the RAID set in a degraded state. With ERC, though, the RAID controller is given time to recover from small read and write errors by rebuilding the specific erroring tracks from the other disks, rather than having to rebuild the entire RAID set because it thinks the disk is bad. Single-sector errors and remaps are pretty common, and without a disk that supports ERC you run the risk of a tiny, easily recoverable error transforming into a system-impacting event.

Enterprise-class drives might be overkill, however. I purchased the device in late April 2012, so at the time Western Digital hadn't yet released its WD Red NAS drives; if they had been available, I would likely have gone with them instead. They offer a good compromise between the reliability and extra firmware features of enterprise-class drives, and the lower cost of consumer-grade gear.