Yes, but all spinny rust kinda sucks compared to SSD. The fastest HDD and a ST2000LM003 differ, yes, but the latter is still at least half as fast as the former.
I too noticed that there's a big fat hole in the WD Red 2.5 lineup, especially annoying as they make a Green 2.5 2TB.
I've got a VM filer here that was originally going to be made up of three different models of low cost 2.5 2TB. Ultimately that got cut to two ... the Toshy MQ01ABB200 and the Sammy ST2000LM003 (now Seacrate). So I've got an array with about a dozen of the Sammies and some of the Toshies, The Toshies are noticeably a bit slower. The Sammies have been very pleasant to work with, no failures that I can recall, which is unusual for a dozen drives that've been running for more than half a year. Actually the remainder of the drives are supposed to be here today.
I can't promise you that they'll be blazing fast or anything like that.
Code:
# camcontrol devlist
<SAMSUNG MZHPU256HCGL-00004 UXM6501Q> at scbus0 target 0 lun 0 (pass0,ada0)
<LSI SAS2X36 0e12> at scbus1 target 8 lun 0 (pass1,ses0)
<ATA ST2000LM003 HN-M 0007> at scbus1 target 9 lun 0 (pass2,da0)
<ATA TOSHIBA MQ01ABB2 0U> at scbus1 target 10 lun 0 (pass3,da1)
<ATA ST2000LM003 HN-M 0007> at scbus1 target 11 lun 0 (pass4,da2)
<ATA ST2000LM003 HN-M 0007> at scbus1 target 12 lun 0 (pass5,da3)
<ATA TOSHIBA MQ01ABB2 0U> at scbus1 target 13 lun 0 (pass6,da4)
<ATA ST2000LM003 HN-M 0007> at scbus1 target 14 lun 0 (pass7,da5)
<ATA ST2000LM003 HN-M 0007> at scbus1 target 15 lun 0 (pass8,da6)
<ATA ST2000LM003 HN-M 0007> at scbus1 target 16 lun 0 (pass9,da7)
<ATA ST2000LM003 HN-M 0007> at scbus1 target 17 lun 0 (pass10,da8)
<ATA ST2000LM003 HN-M 0007> at scbus1 target 18 lun 0 (pass11,da9)
<ATA ST2000LM003 HN-M 0005> at scbus1 target 19 lun 0 (pass12,da10)
<ATA ST2000LM003 HN-M 0005> at scbus1 target 20 lun 0 (pass13,da11)
<ATA ST2000LM003 HN-M 0007> at scbus1 target 21 lun 0 (pass14,da12)
<ATA ST2000LM003 HN-M 0007> at scbus1 target 22 lun 0 (pass15,da13)
<SATA SSD S9FM01.6> at scbus10 target 0 lun 0 (pass16,ada1)
<SATA SSD S9FM01.6> at scbus11 target 0 lun 0 (pass17,ada2)
# dd if=/dev/da13 of=/dev/null bs=1048576 count=1024
1024+0 records in
1024+0 records out
1073741824 bytes transferred in 8.471599 secs (126746065 bytes/sec)
At ~100-120MB/sec they perform adequately for sequential read. The platform above is a PoC VM filer that uses mirror vdevs, the design calls for seven three-way mirrors, which gives lots of random read capacity with less write capacity. 48TB of raw disk (seven vdevs plus three spare drives) will be used to provide a 14TB pool, of which we won't use more than 7TB. SSD would be a better choice in a year or two I think, but right now the costs aren't favoring that. The platform is suitable for our needs. I'm intending to bump it up from 64GB to 128GB RAM and then add some more L2ARC. It'll be very pleasant, I think.
I've been abusing laptop-grade HDD's in servers for a number of years now, and as long as you're not being stupid abusive, they don't fail much more frequently than the hotter running SAS drives. The difference is that when a 500GB SATA 2.5 drive fails, you can run down to Best Buy and get a replacement for $45. At that price, who cares. Average survivability of the Seagate Momentus XT (hybrid 500GB) drives seems to be around four years, we just saw two of those that we deployed in 2010(?) fail in the past few months.
