Long time no post – how about some SATA disk benchmarks?

It’s been a long time since I’ve posted to my fledgling blog, mainly because I’m working … aLOT.

ALOT2

 

… And I really do care about this blog a lot. And the two people that have left comments made me care just a little more. Unfortunately I just don’t get a chance to post all the meh’s I find while working and playing with technology. However, posting some benchmarks I did for fun is easy so here goes.

First, I have a little server sitting under my desk at work that I use a VM lab. For posterity, here are the specs.
AMD Athlon x3 720 (with 4th core unlocked)
24GB DDRs
2 1Gbs NICs
80GB SATA OS
1TB RAID 0 array with 4x 250GB SATA disks
Windows Server 2012 Datacenter x64

Since it’s a lab I don’t really care about backup and disk failure so RAID 0, also the built in RAID doesn’t support RAID 5. Anyway, I had 4x 500GB SATA drives, some older Seagate drives that are roughly 5 years old. They were slow, and one squeaked. I did my best to tweak stuff to get the most IO out of them, but my VM’s felt like they were on a single IDE drive. I also tried RAID 10 but the speed on that was far worse than RAID 0. Of course you’ll get better read speeds with this setup in RAID 0, but the writes will suffer. ?In my real world touchy feely analysis, RAID 0 felt faster.

Thankfully work bought a bunch of new servers with 250GB SATA drives in them that we don’t use, we upgraded to SAS drives and now we have a box full of these 250GB guys. So I grabbed 4 of them and built a new RAID 0 array. Below are the results of my quick benches with Crystal DiskMark. It’s not a very scientific test, I just wanted to verify that one RAID was faster than the other (even though technically I know it should be) and the new disks were faster than the older even though they’re all SATAII (300MB/s). I was right, RAID 0 is faster than RAID 10 in this setup and the new disks are faster than the old. Real time feely performance is much better as well. RDP is snappier, VM’s boot faster, update faster, etc. Plus my disk queue is incredibly low considering this is SATA. Hovers around 1 or less. Used to get up to 10 🙁

Conclusion: New drives are faster, RAID 0 is faster than RAID 10. So if you’re lab VM host is suffering from poor IO, go get some cheap new SATA drives and speed that thing up.

*I didn’t test RAID 10 with the new disks because I wanted more disk space and I was in a hurry.
250GB disk model – Dell “Enterprise Class” Western Digital manufactured WD2502ABYS-18B7A0
500GB disk model – Seagate ST3500641AS

———————————————————————–
New Drives – 4x 250GB HDD RAID 0 – 50MB Test
———————————————————————–

Sequential Read : 403.166 MB/s
Sequential Write : 212.848 MB/s
Random Read 512KB : 119.966 MB/s
Random Write 512KB : 202.434 MB/s
Random Read 4KB (QD=1) : 4.972 MB/s [ 1213.9 IOPS]
Random Write 4KB (QD=1) : 9.415 MB/s [ 2298.5 IOPS]
Random Read 4KB (QD=32) : 11.597 MB/s [ 2831.3 IOPS]
Random Write 4KB (QD=32) : 14.956 MB/s [ 3651.3 IOPS]

Test : 50 MB [E: 0.0% (0.2/930.9 GB)] (x1)

———————————————————————–
New Drives – 4x 250GB HDD RAID 0 – 1GB Test
———————————————————————–

Sequential Read : 433.057 MB/s
Sequential Write : 433.415 MB/s
Random Read 512KB : 54.181 MB/s
Random Write 512KB : 116.335 MB/s
Random Read 4KB (QD=1) : 0.718 MB/s [ 175.2 IOPS]
Random Write 4KB (QD=1) : 4.874 MB/s [ 1189.9 IOPS]
Random Read 4KB (QD=32) : 4.609 MB/s [ 1125.2 IOPS]
Random Write 4KB (QD=32) : 7.778 MB/s [ 1898.8 IOPS]

Test : 1000 MB [E: 0.0% (0.2/930.9 GB)] (x1)

———————————————————————–
4x 500GB RAID 10 ::::::::::::::::::::::::::::::::::::::::::::::::::::::
———————————————————————–

Old drives – RAID 10 x4 500GB HDD – 50MBTest
———————————————————————–

Sequential Read : 86.702 MB/s
Sequential Write : 63.977 MB/s
Random Read 512KB : 43.530 MB/s
Random Write 512KB : 40.987 MB/s
Random Read 4KB (QD=1) : 1.249 MB/s [ 305.0 IOPS]
Random Write 4KB (QD=1) : 1.173 MB/s [ 286.4 IOPS]
Random Read 4KB (QD=32) : 3.571 MB/s [ 871.8 IOPS]
Random Write 4KB (QD=32) : 1.371 MB/s [ 334.8 IOPS]

Test : 50 MB [F: 3.3% (30.6/931.3 GB)] (x1)
Date : 2013/07/19 17:51:15
OS : Windows Server 2012 Datacenter Edition (Full installation) [6.2 Build 9200] (x64)

———————————————————————–
Old drives – RAID 10 x4 500GB HDD – 1GB Test
———————————————————————–

Sequential Read : 94.722 MB/s
Sequential Write : 79.879 MB/s
Random Read 512KB : 34.521 MB/s
Random Write 512KB : 38.625 MB/s
Random Read 4KB (QD=1) : 0.506 MB/s [ 123.6 IOPS]
Random Write 4KB (QD=1) : 1.100 MB/s [ 268.5 IOPS]
Random Read 4KB (QD=32) : 2.300 MB/s [ 561.5 IOPS]
Random Write 4KB (QD=32) : 1.214 MB/s [ 296.4 IOPS]

Test : 1000 MB [F: 3.3% (30.8/931.3 GB)] (x1)

———————————————————————–
Old drives – RAID 0 x4 500GB HDD – 50MB Test
———————————————————————–

Sequential Read : 208.395 MB/s
Sequential Write : 150.298 MB/s
Random Read 512KB : 55.850 MB/s
Random Write 512KB : 136.771 MB/s
Random Read 4KB (QD=1) : 1.700 MB/s [ 415.0 IOPS]
Random Write 4KB (QD=1) : 2.786 MB/s [ 680.2 IOPS]
Random Read 4KB (QD=32) : 6.128 MB/s [ 1496.2 IOPS]
Random Write 4KB (QD=32) : 3.749 MB/s [ 915.4 IOPS]

Test : 50 MB [E: 0.0% (0.2/1862.6 GB)] (x1)

———————————————————————–
Old drives – RAID 0 x4 500GB HDD – 1GB Test
———————————————————————–

Sequential Read : 202.565 MB/s
Sequential Write : 181.007 MB/s
Random Read 512KB : 43.560 MB/s
Random Write 512KB : 42.235 MB/s
Random Read 4KB (QD=1) : 0.666 MB/s [ 162.5 IOPS]
Random Write 4KB (QD=1) : 1.530 MB/s [ 373.6 IOPS]
Random Read 4KB (QD=32) : 3.274 MB/s [ 799.4 IOPS]
Random Write 4KB (QD=32) : 2.204 MB/s [ 538.0 IOPS]

Test : 1000 MB [E: 0.0% (0.2/1862.6 GB)] (x1)

2 Comments

 Add your comment
  1. Just found your site. Just in case you want to see one of the things I’m working on.

    micron-p320h-hhhl-700gb-pcie-enterprise-ssd-review-vertical-integration-high-iops-and-absurd-endurance

  2. That’s awesome. How do I get one? I’ll review it! 😛 We’re in the process of purchasing a distributed SSD SAN at work and I don’t think the entire 40 disk array does that many IOPs. But it does a LOT more than our 15k SAS array 🙂

Leave a Comment

Your email address will not be published.

This site uses Akismet to reduce spam. Learn how your comment data is processed.