SSD All The Things

After some grueling IO testing on 7200rpm disks, I got my hands on some shiny new Samsung 840 SSDs and wanted to share the performance results in similar fashion.

Dell r630, E5-2630 v3 @ 2.40GHz (16 cores), 256GB RAM, Samsung 840 EV SSD 960GB.

Testing was sysbench, fileio-rndrw, async+direct io, 16 threads, 5.5TB RAID-6 using XFS mounted with noatime,inode64 and using the deadline scheduler.

Write Policy Read Policy xfs options Transfer/s Requests/s Avg/Request 95%/Request
WB ADRA sunit=16, swidth=576 blks 357.31Mb/sec 22868.08 0.17ms 0.70ms
WT ADRA sunit=16, swidth=576 blks 459.27Mb/sec 29393.37 0.14ms 0.53ms
WT NORA sunit=16, swidth=576 blks 448.07Mb/sec 28676.79 0.14ms 0.50ms
WT ADRA sunit=16, swidth=384 blks 450.69Mb/sec 28843.85 0.16ms 0.51ms

WT = Write Through, WB = Write Back, ADRA = Adaptive Read Ahead; NORA = No Read Ahead

It’s a no brainer that SSDs are faster. For our testing, they were 7x faster than the previous 7200rpm tests and 4x faster than 15k platter in RAID-10 (15k results not published).

Edit: Clarified policy information and the speed differences from 7200/15k. Thanks Daniël van Eeden.


5 thoughts on “SSD All The Things

  1. Etienne

    Thank you for sharing. I’m wondering how you plug your SSD in a R630, are you using a RAID controler from DELL ? If so how do you manage TRIM commands which aren’t supported ?

    1. jeremytinley Post author


      We are using the PERC H730P controller. You’re right, we do lose TRIM support. Our production will end up using the Dell SSDs instead of the Samsung. The Samsung drives were less expensive and made for a simple proof of concept test to verify the IOPS capacity as it relates to our workload. The Dell (re)branded ones should have a higher endurance rating.

  2. Daniël van Eeden

    I guessed WPol = write policy and RPol = read policy and NORA = no read ahead and ADRA = adaptive read ahead?

    In your previous blog you reached 4126 ops, now 29393. That’s a more that 7x improvement, not a 4x improvement. And the same is true for MB/s.

    1. jeremytinley Post author

      Correct, I’ll edit and clarify. I was posting it rather abruptly.

      I reached 4129 on a 7200rpm disk. The unpublished 15k rpm were ~7000, which we use today in production, was the comparison I was making at bottom.

  3. Raju

    Thank you for sharing the results, Can you please provide how many 7200rpm drives were in the system, also how many 15krpms drives in the system?


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s