Hard Disk test and iops test. It is very rough. It is tested in the win8pe32 environment. It is not a professional tool, but it is enough... Two Disks (two disks, and two disks). AHCI is enabled in the single disk test interface mode.
Sata1.0 ===== motherboard: Aopen/i945GTm-VHL === cpu is the T7200 (dual core) 650 of the notebook) this. width = 650; "title =" sata1.0 "alt =" 230940230.png" src = "http://www.bkjia.com/uploads/allimg/131229/120J35054-0.png"/>
===
650) this. width = 650; "title =" sata1.0atto "alt =" 231100633.png" src = "http://www.bkjia.com/uploads/allimg/131229/120J33101-1.png"/>
It is obviously a bottleneck caused by sata1.0, so the motherboard MUST be changed
Sata2.0 === single and dual disk raid0 === sata2.0 tested hardware: xeon3040 + G43T + ddr2-2gx1
So far, a single mechanical hard disk cannot exceed the M rate of sata2.0. Therefore, if you do not consider sata3.0, it is cost-effective. Ignore this!
=== Sata2.0 single disk read
650) this. width = 650; "title =" sata2.0read "alt =" 232047605.png" src = "http://www.bkjia.com/uploads/allimg/131229/120J32325-2.png"/>
= Sata2.0 single disk write 650) this. width = 650; "title =" sata2.0write "alt =" 232236685.png" src = "http://www.bkjia.com/uploads/allimg/131229/120J310P-3.png"/>
=== Sata2.0 single disk ATTO Software Testing
650) this. width = 650; "title =" sata2.0atto "alt =" 232744484.png" src = "http://www.bkjia.com/uploads/allimg/131229/120J34112-4.png"/>
= Sata2.0-raid0 read 650) this. width = 650; "title =" sata2.0raid0 "alt =" 232515636.png" src = "http://www.bkjia.com/uploads/allimg/131229/120J34E9-5.png"/>
= Saa2.0-raid0 write 650) this. width = 650; "title =" sata2.0raid0write "alt =" 233524806.png" src = "http://www.bkjia.com/uploads/allimg/131229/120J33H8-6.png"/>
=== Sata2.0-raid0 "ATTO
650) this. width = 650; "title =" raid0atto "alt =" 232941962.png" src = "http://www.bkjia.com/uploads/allimg/131229/120J3KK-7.png"/>
Conclusion: Although the performance of raid0 has doubled, the cpu usage has also doubled. Do not forget that the maximum value of a gigabit Nic is 125 MB/second. The speed of a single disk is enough to feed the gigabit network card. raid0 also has the risk of bad disks, so there is no need for raid0.
IOPS test section ===separate single disk and dual disk RAID 0 ==== this part is also under sata2.0
= Iops sata2.0 single disk read 650) this. width = 650; "title =" iopsread "alt =" 235455475.png" src = "http://www.bkjia.com/uploads/allimg/131229/120J33C9-8.png"/>
===
650) this. width = 650; "title =" iopsexread "alt =" 235542512.png" src = "http://www.bkjia.com/uploads/allimg/131229/120J323C-9.png"/>
=== Read iops sata2.0-raid0
650) this. width = 650; "title =" iopsraid0read "alt =" 41059492.png" src = "http://www.bkjia.com/uploads/allimg/131229/120J36492-10.png"/>
===
650) this. width = 650; "title =" iopsraid0exread "alt =" 000148656.png" src = "http://www.bkjia.com/uploads/allimg/131229/120J32G7-11.png"/>
= Iops, sata2.0 single disk, write 650) this. width = 650; "title =" iopswrite "alt =" 235805140.png" src = "http://www.bkjia.com/uploads/allimg/131229/120J3I48-12.png"/>
===
650) this. width = 650; "title =" iopsexwrite "alt =" 235905820.png" src = "http://www.bkjia.com/uploads/allimg/131229/120J324Y-13.png"/>
= Iops sata2.0-raid0 write 650) this. width = 650; "title =" iopsraid0write "alt =" 000314581.png" src = "http://www.bkjia.com/uploads/allimg/131229/120J36003-14.png"/>
===
650) this. width = 650; "title =" raid0exwrite "alt =" 002219873.png" src = "http://www.bkjia.com/uploads/allimg/131229/120J33O2-15.png"/>
Iops Summary: in fact, we are quite disappointed to see the data. The iops of raid0 is not much higher than the iops of a single disk. This is also a big reason for deciding not to use raid0. What about raid5... Against the low-cost principle, the purchase of RAID cards deviated from the original intention.
!!!
This article is from the "DIY virtualization environment" blog. For more information, contact the author!