Announcement

Collapse
No announcement yet.

IO_uring Passthrough For KVM Guests Is Yielding Fantastic Results

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • IO_uring Passthrough For KVM Guests Is Yielding Fantastic Results

    Phoronix: IO_uring Passthrough For KVM Guests Is Yielding Fantastic Results

    At the Linux Foundation's KVM Forum 2020 this week, Stefano Garzarella of Red Hat presented on the possibility of speeding up virtual machine I/O performance by sharing the host's IO_uring queues with guests. In doing so the virtual machines have near bare-metal performance with the proposed patches...

    http://www.phoronix.com/scan.php?pag...through-LF2020

  • #2
    One of the best innovation in Linux. It's positive that 5.10 will be the next LTs kernel.

    Comment


    • #3
      Does it mean OSes in VirtualBox will load faster?

      Comment


      • #4
        Is IO_uring used always by default?

        Comment


        • #5
          Originally posted by cl333r View Post
          Does it mean OSes in VirtualBox will load faster?
          KVM = Kernel Virtual Machine. The userland part you (could) know is qemu.

          If VirtualBox is going to support io_uring passthrough, it might.
          I wouldn't place my bets on anything soon there.
          I'm guessing QEMU would be a logical first contender to pick it up. But even then, i'm also guessing there needs to be some guest machine support for this.

          In other terms. Nice new tech but you likely won't be using it anytime soon.

          Comment


          • #6
            Originally posted by markg85 View Post
            KVM = Kernel Virtual Machine. The userland part you (could) know is qemu.
            In VirtualBox you can choose which virtualization to use. KVM is one among a few others.

            Comment


            • #7
              WHOAH !!

              I got nothin' else !

              Comment


              • #8
                So IO_uring is basically the name for asynchronous disk I/O?
                Do applications call some specific APIs like fwrite_async instead of fwrite? Does the file system need to support it?

                Why has Linux gotten IO_uring just lately? Windows had async disk I/O almost forever?

                Comment


                • #9
                  Originally posted by uid313 View Post
                  So IO_uring is basically the name for asynchronous disk I/O?
                  Do applications call some specific APIs like fwrite_async instead of fwrite? Does the file system need to support it?

                  Why has Linux gotten IO_uring just lately? Windows had async disk I/O almost forever?
                  LInux had async io api 'aio' for a lot of time too. My understanding is that io_uring is a more optimized one.
                  Last edited by Random_Jerk; 01 November 2020, 04:12 PM.

                  Comment


                  • #10
                    uid313

                    https://lwn.net/Articles/810414/

                    LWN Article from January 2020.

                    Explains quite a bit of the background information.

                    And the comparison to windows is... broccoli to apples.


                    Comment

                    Working...
                    X