Benchmarking individual drives

One of the aspects of RAID’ing drives of different makes, models, and performance is that your slowest drive will be your bottleneck.
To aid in identifying your slowest drive, below is a benchmark script.

  1. Download: http://download.openegg.org/temp/Simple-IO-Test.zip
  2. There are two example test scripts. The first one benchmarks each drive at time and the other does it in parallel.
    Do the individual test first and then the parallel one after.
Be Sociable, Share!

Revisions

4 Responses to “Benchmarking individual drives”

  1. Neil July 13, 2012 at 4:34 AM #

    When you say bottleneck do you mean the slowest disk will perminantly slow down the entire pool? Or will the pool only suffer slower performance when flexriad happens to be reading or writing to the slowest disk?

  2. Gabriel August 24, 2012 at 8:42 AM #

    How do I run this test on a Windows 7 x64 machine?

  3. Jeff February 6, 2013 at 9:49 PM #

    I noticed over several months of usage, my drives were going to sleep. It would take a very long time for them to wake on my first access of my drive pool in the morning…

    Then I started reading about Windows 7 performance tuning and power settings. I have never done it before, but a desktop machine first installed with the OS will have a setting to turn off drives after 20 minutes. This can cause FlexRaid to be very slow until all the drives in your pool wake from sleep settings. I would sometimes experience several minutes of delays when I would first try to access the drive pool; completely unacceptable performance no matter what the energy savings, for a server storage array.

    It was as easy as going into the control panel, Hardware and Sound, Power Settings, choosing High Performace Settings option, and then Change Plan Settings for that selection, and go to Advanced Settings, Hard Disk and setting Turn Off Hard Disk to NEVER.

    Now I experience ENTIRELY DIFFERENT PERFORMANCE from FlexRaid and I would never have known this from the FlexRaid documentation. I assume the developer is not entirely involved with Windows and has not experienced this complaint. Either way, make sure you understand this if you host FlexRaid on Windows.

  4. Jeff February 6, 2013 at 10:17 PM #

    With above settings my FlexRaid pool is very responsive, very quick, even with both 5400 and 7200 RPM drives included.
    Not sure it makes a huge difference in user experience, except in transfer times, and that will be more oriented around the device being accessed than the “slowest drive in the pool”.

    That said, I am interested to replace all my drives with the latest 3TB Seagate drives (purchased individually from separate retailers over a years time to avoid bad batches of drives) and eliminate and sell off all my old drives!

    That would I believe, optimize the array for best speed and performance:

    Min # of disk drives X best RPM speed X quiet operation X modern platter capacity === best pool array performance and noise

    Now off to Ebay to sell, sell, sell…

Leave a Reply

one × three =