Results 1 to 6 of 6

Thread: QED: A New, High Performance QEMU Disk Format

  1. #1
    Join Date
    Jan 2007
    Posts
    14,536

    Default QED: A New, High Performance QEMU Disk Format

    Phoronix: QED: A New, High Performance QEMU Disk Format

    Linux-KVM mentions QED, the new QEMU Enhanced Disk format. This new disk format for QEMU/KVM is designed to be much faster than QCOW2 and other existing disk formats available to virtualization users...

    http://www.phoronix.com/vr.php?view=OTY0MQ

  2. #2
    Join Date
    Oct 2006
    Location
    Israel
    Posts
    572

    Default

    It should be interesting to benchmark qed against qcow and raw.

    - Gilboa
    DEV: Intel S2600C0, 2xE52658V2, 32GB, 4x2TB, GTX680, F20/x86_64, Dell U2711.
    SRV: Intel S5520SC, 2xX5680, 36GB, 4x2TB, GTX550, F20/x86_64, Dell U2412..
    BACK: Tyan Tempest i5400XT, 2xE5335, 8GB, 3x1.5TB, 9800GTX, F20/x86-64.
    LAP: ASUS N56VJ, i7-3630QM, 16GB, 1TB, 635M, F20/x86_64.

  3. #3
    Join Date
    Jun 2008
    Location
    Perth, Scotland
    Posts
    435

    Default

    Isn't raw access to a real disk always faster?

  4. #4

    Default

    Quote Originally Posted by Chewi View Post
    Isn't raw access to a real disk always faster?
    Does RAW provide access to TRIM?

  5. #5
    Join Date
    Jun 2008
    Location
    Perth, Scotland
    Posts
    435

    Default

    Quote Originally Posted by Shining Arcanine View Post
    Does RAW provide access to TRIM?
    I think that's irrelevant if you're using a real disk.

  6. #6
    Join Date
    Oct 2006
    Location
    Israel
    Posts
    572

    Default

    trim support doesn't necessarily require non-raw file system.
    Actually, I'm far from being an expert in this field, but I would imagine that implementing virtualized hardware w/ trim support is somewhat easier on raw-FS-partition, as qemu can simply pass the trim request (w/ address translation) from the guest to the underlying hardware.

    ... but as I as I said, I'm guessing

    As for performance, I would guess that raw will have the upper hand given the same optimization on the qemu side (E.g. write combination, etc). But given the fact that most of the effort is spent on optimizing advanced image types (qcow, etc), it's entirely possible that raw will not win the benchmark(s).

    - Gilboa
    DEV: Intel S2600C0, 2xE52658V2, 32GB, 4x2TB, GTX680, F20/x86_64, Dell U2711.
    SRV: Intel S5520SC, 2xX5680, 36GB, 4x2TB, GTX550, F20/x86_64, Dell U2412..
    BACK: Tyan Tempest i5400XT, 2xE5335, 8GB, 3x1.5TB, 9800GTX, F20/x86-64.
    LAP: ASUS N56VJ, i7-3630QM, 16GB, 1TB, 635M, F20/x86_64.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •