Announcement

Collapse
No announcement yet.

Intel Haswell Linux Virtualization: KVM vs. Xen vs. VirtualBox

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

  • #11
    So If I wanted to run a virtualized instance of Windows for windows-only software, what would be the best idea? I'm thinking vmware workstation, but I'm worried since fedora updates their kernels semi-frequently and things might break with the modules and I'll really need this machine to be reliable.
    Then again I could just put off updating the kernel for a bit until vmware has an update. Or even use a 3rd party patch.

    Comment


    • #12
      Originally posted by ownagefool View Post
      Xen and KVM are bare metal hypervisors, where virtual box is a hosted hypervisor. Doing your virtualization at application level is always going to be slower than at kernel level unless somethings really wrong.
      not only that. they are paravirtualization. the hardware doesn't have to be fully emulated. that's why I can run two guests on KVM or Hyper-V and the processor and the hard drive are not going crazy.

      Comment


      • #13
        Originally posted by garegin View Post
        not only that. they are paravirtualization. the hardware doesn't have to be fully emulated. that's why I can run two guests on KVM or Hyper-V and the processor and the hard drive are not going crazy.
        That is partly true, only Xen can run full PV guests. In the benchmarks the Xen guest is installed as a HVM machine using PV drivers, similar to KVM. It would be interesting to see i a real PV guest installation in Xen would make a difference. My personal experience is that real PV guests are snappier, but I know they suffer in some benchmarks.
        Even more interesting would be to see tests on the upcoming PVHVM mode in Xen 4.3 [URL="http://wiki.xen.org/wiki/Xen_Overview#PV_in_an_HVM_Container_.28PVH.29_-_New_in_Xen_4.3"]

        Comment


        • #14
          Originally posted by LasseKongo View Post
          That is partly true, only Xen can run full PV guests. In the benchmarks the Xen guest is installed as a HVM machine using PV drivers, similar to KVM. It would be interesting to see i a real PV guest installation in Xen would make a difference. My personal experience is that real PV guests are snappier, but I know they suffer in some benchmarks.
          Even more interesting would be to see tests on the upcoming PVHVM mode in Xen 4.3 [URL="http://wiki.xen.org/wiki/Xen_Overview#PV_in_an_HVM_Container_.28PVH.29_-_New_in_Xen_4.3"]
          Red Hat installed inside Xen should be fully paravirtualized - the installer will detect the Xen kernel and install the appropriate drivers.

          Comment


          • #15
            Originally posted by garegin View Post
            not only that. they are paravirtualization. the hardware doesn't have to be fully emulated. that's why I can run two guests on KVM or Hyper-V and the processor and the hard drive are not going crazy.
            Paravirtualized drivers are also supported on Virtual box. Xen is the only one that uses a paravirtualized kernel - the others use standard kernels, and rely on hardware extensions like VT-X for speed.

            Comment


            • #16
              Originally posted by ownagefool View Post
              Xen and KVM are bare metal hypervisors, where virtual box is a hosted hypervisor. Doing your virtualization at application level is always going to be slower than at kernel level unless somethings really wrong.
              Xen vs KVM arguments and marketing, both sides argue that their product is"bare metal"while the other is "hosted". Take it all with a grain of salt. In reality, PC emulation for both is done in user space using qemu. Qemu is not running in the kernel. Some drivers are paravirtualized but a lot of stuff (BIOS etc) is emulated with qemu. Both have a hypervisor, whether this is a full Linux kernel or some custom small kernel doesn't really matter that much. Both support VT-X so a paravirtualized kernel isn't necessary any more. If I recall correctly from benchmarks there's no difference between paravirtualized Xen kernel and hardware accelerated at this point. Drivers are another matter, paravirtualized drivers are still necessary to boost performance, but those are available on all platforms including Virtual box. Virtual Box supports VT-X as well, so performance isn't really that different. Check this previous comparison http://www.phoronix.com/scan.php?pag...0_xenkvm&num=2

              Comment


              • #17
                Originally posted by n3wu53r View Post
                So If I wanted to run a virtualized instance of Windows for windows-only software, what would be the best idea? I'm thinking vmware workstation, but I'm worried since fedora updates their kernels semi-frequently and things might break with the modules and I'll really need this machine to be reliable.
                Then again I could just put off updating the kernel for a bit until vmware has an update. Or even use a 3rd party patch.
                I've used both and in the end Virtual Box wins. Virtual Box will do everything that you want, plus it supports easy snapshots of you ever want to try something out then roll back the whole system. VMware workstation has closed source kernel drivers with the usual problems upgrading.

                Comment


                • #18
                  These benchmark results are exactly what I expected. These days KVM performs damn well, and I use it for just about everything. We used to use Xen a lot in the lab, but many times over the years I've had issues when running newer guests. KVM has always worked well, and has picked up a ton of nice features recently.

                  KVM also has snappshotting, provided your using the right disk image (such as qcow2). Snapshotting is not yet available in the Virt-Manager GUI, which I'm guessing most people would use. Doing a snapshot via virsh is dead simple though.

                  Comment


                  • #19
                    Originally posted by benmoran View Post
                    These benchmark results are exactly what I expected. These days KVM performs damn well, and I use it for just about everything. We used to use Xen a lot in the lab, but many times over the years I've had issues when running newer guests. KVM has always worked well, and has picked up a ton of nice features recently.

                    KVM also has snappshotting, provided your using the right disk image (such as qcow2). Snapshotting is not yet available in the Virt-Manager GUI, which I'm guessing most people would use. Doing a snapshot via virsh is dead simple though.
                    Isn't there a large performance hit with qcow2 though? Eg http://michaelmk.blogspot.co.uk/2012...mark.html?m=1#! shows some benchmarks at only 10% of LVM performance.

                    Comment


                    • #20
                      Originally posted by n3wu53r View Post
                      So If I wanted to run a virtualized instance of Windows for windows-only software, what would be the best idea? I'm thinking vmware workstation, but I'm worried since fedora updates their kernels semi-frequently and things might break with the modules and I'll really need this machine to be reliable.
                      Then again I could just put off updating the kernel for a bit until vmware has an update. Or even use a 3rd party patch.
                      I use a KVM for exactly that and it's fine. But I don't do any 3D stuff: if you do 3D stuff and you need passthrough acceleration VBox is really the only option you have ATM. (FWIW, the Fedora kernel and virt devs are unified in viewing VBox as a terrible, terrible piece of software; they obviously think KVM is the best thing ever, but it's not just competition-syndrome, they think VMware and Xen are perfectly fine code, it's just VBox they think is really terrible).

                      Comment

                      Working...
                      X