Creative Communities of the World Forums

The peer to peer support community for media production professionals.

  • Tom Bridges

    November 28, 2005 at 1:18 pm

    [Gary Taylor] “Did you give any thought the implementing one of these? I am wondering if the ASIC would compensate for the overhead for RAID 6 writes.”

    Haven’t tried RAID 6 yet. Not entirely sure how it works, actually. I believe the idea is that there are two sets of parity, so any two drives can fail? Something like that. In any case, we’re not willing to sacrifice space and performance to that extent: one hot spare is enough. If I get time, I’ll do the tests.

    [David Cherniack] “A complaint I’ve heard about the Infotrend raids is the variation – “all over the map” is how one dealer who evaluated the system put it. The average is still good, but not if it dips to 200MB/S in the valleys. So please know how far it dipped as well as at 80% full when you can do the test.”

    Good point. The lowest I’ve got out of the RAID 50 through Quickbench is 425MB/s. So still substantial variation but definitely in the safe zone. I’m experimenting with different stripe sizes at the moment to see if that affects performance/variation.

    Best,

    Tom

    Split Image
    http://www.split-image.co.uk

  • Yves De muyter

    November 28, 2005 at 1:55 pm

    Please fill up your array up to 70-80%. There is no sense in testing an empty array, we all know almost any array, even a single disk, is fast when empty.

    If needed, you can fill your array easily by duplicating media that is on it.

    -Yves

  • David Cherniack

    November 28, 2005 at 3:08 pm

    [Tom Bridges] “The lowest I’ve got out of the RAID 50 through Quickbench is 425MB/s.”

    I’ll be interested to see the test with 80%.

    David
    AllinOneFilms.com

  • Tom Bridges

    November 29, 2005 at 12:30 pm

    So I’ve filled the RAID up to over 80%. For the reads, it’s benching between 400MB/s and 700MB/s whilst writes are between 600-640MB/s. I’ll fill it to the brim and see what happens.

    Hope this helps,

    Tom

    Split Image
    http://www.split-image.co.uk

  • David Cherniack

    November 29, 2005 at 1:13 pm

    Thanks for doing these tests, Tom.

    David
    AllinOneFilms.com

  • Tom Bridges

    November 29, 2005 at 3:30 pm

    No problem David. It’s looking like a pretty robust system: when completely full, the disk speeds aren’t seriously affected. Minimum reads is at 340MB/s (max 720MB/s!) and writes slightly higher (640-660MB/s). Which is a bit weird, really. I look forward to getting my hands on some Blackmagic hardware so I can run the Speed Test utility. As I mentioned, this is all done with Quickbench. The precise RAID configuration is two RAID 5 volumes (with hot spare), which is then software striped using Disk Utility. I’m going to run a redundancy test when I have a moment.

    Best wishes,

    Tom

    Split Image
    http://www.split-image.co.uk

  • David Cherniack

    November 29, 2005 at 8:16 pm

    [Tom Bridges] “Which is a bit weird, really. I look forward to getting my hands on some Blackmagic hardware so I can run the Speed Test utility.”

    It seems to me that it’s very weird to get faster writes than reads. Too bad you’re not on a PC or I’d suggest Diskperf.

    David
    AllinOneFilms.com

  • George Kalishkin

    March 16, 2006 at 1:06 pm

    can you test with Blackmagic’s Speed Test?

    how many cache memory there is in raid controler?

Page 2 of 2

We use anonymous cookies to give you the best experience we can.
Our Privacy policy | GDPR Policy