I compared I/O performance of three drives:
- Kingston SSDNow V300 Series SV300S37A/120G 2.5" 120GB SATA III SSD - just because it was already there.
- Single hard drive - 2861589GB Seagate Barracuda 7200.14 (AF) ST3000DM001-1E6166
- A Microsoft Windows 8.1 Storage Space with two of the above hard drives configured as mirror.
MB/s | IOPS | MB/s | IOPS | MB/s | IOPS | Delta | |
Sequential Read: | 446.836 | 140.259 | 181.808 | 22.85% | |||
Sequential Write: | 144.154 | 137.572 | 125.443 | -8.82% | |||
Random Read 512KB: | 404.102 | 50.204 | 39.758 | -20.81% | |||
Random Write 512KB: | 144.021 | 82.208 | 66.927 | -18.59% | |||
Random Read 4KB (QD=1): | 22.666 | 5533.8 | 0.386 | 94.3 | 0.449 | 109.6 | 14.03% |
Random Write 4KB (QD=1): | 108.644 | 26524.5 | 6.93 | 1692 | 0.382 | 93.2 | -94.49% |
Random Read 4KB (QD=32): | 116.621 | 28471.9 | 0.468 | 114.3 | 0.456 | 111.2 | -2.56% |
Random Write 4KB (QD=32): | 141.824 | 34624.9 | 0.983 | 239.9 | 0.384 | 93.7 | -60.94% |
SSD | 1x Barracuda 7200.14 | 2x Barracuda 7200.14 |
Transition to storage spaces mirror moderately improves sequential read speed. Random write performance however is dramatically deteriorated. For a workstation use scenario this makes storage spaces unacceptable for me. They may still be relevant for servers.
So I will be sticking with two single hard drives. Maybe will use synchtoy to backup one drive to another.