Linux KVM vs. VirtualBox 4.0 Virtualization Benchmarks

[phoronix]

Phoronix: Linux KVM vs. VirtualBox 4.0 Virtualization Benchmarks

Oracle's been vigorously working on their VM VirtualBox 4.0 software and in just the past week they have delivered two public betas that bring a number of new features. Among the changes there is support for Intel HD audio / ICH9 to guest VMs, the concept of extension packs, user-interface improvements, support for limiting a virtual machine's CPU time and I/O bandwidth, 3D acceleration fixes for guests, and a great number of bug-fixes. How though is this updated Oracle/Sun virtualization platform comparing to the older VirtualBox 3.2 release and that of the upstream Linux KVM (the Kernel-based Virtual Machine) that most Linux distributions rely upon? Here are a number of benchmarks that seek to answer this very question.

http://www.phoronix.com/vr.php?view=15549

[birdie]

we allowed the guests access to all 12 processor threads

and that's why the whole test makes no sense. Virtualization is known to work badly with "virtual" "emulated" CPUs. So, next time I'd like to see something a bit more realistic, like HT support disabled in BIOS, host and guest limited to say 4 CPUs (host via kernel parameters).

What this test has shown is that no virtualization technology (amongst tested) scales well with the sheer number of emulated CPUs.

Besides it's kinda sad you haven't included VMWare WorkStation/Player 7.1.3.

[deanjo]

Besides it's kinda sad you haven't included VMWare WorkStation/Player 7.1.3.

Wouldn't mind seeing parallels thrown into the mix as well.

[popper]

and that's why the whole test makes no sense. Virtualization is known to work badly with "virtual" "emulated" CPUs. So, next time I'd like to see something a bit more realistic, like HT support disabled in BIOS, host and guest limited to say 4 CPUs (host via kernel parameters).

What this test has shown is that no virtualization technology (amongst tested) scales well with the sheer number of emulated CPUs.

Besides it's kinda sad you haven't included VMWare WorkStation/Player 7.1.3.

its Not disabling Hyper-threading (as in using the spare cycles of the real CPU's to get up to 52% more throughput in these virtual CPU thread's) But rather disabling 'Turbo Boost' to remove the freaky and unpredictable nature that is TB, as can be seen If you run a simple ./checkasm --bench after pulling the x264 git source for instance..

[frantaylor]

Besides it's kinda sad you haven't included VMWare WorkStation/Player 7.1.3.

What's sad is that you did not read the EULAs that you agreed to:

VMware Player and Workstation EULA:

"You may use the Software to conduct internal performance testing and benchmarking studies, the results of which you (and not unauthorized third parties) may publish or publicly disseminate; provided that VMware has reviewed and approved of the methodology, assumptions and other parameters of the study."

[droidhacker]

I would like to know more about the loopback test. Where did the data come from and where did it go to? Reason for asking is that the same thing that caused virtualbox to cheat sqlite and other disk tests could have been affecting the loopback test depending on the way the test worked.

In other words, the only place where virtualbox had a commanding lead over kvm that was not NECESSARILY a result of cheating, may not be trustworthy.

[Harri]

and that's why the whole test makes no sense. Virtualization is known to work badly with "virtual" "emulated" CPUs. So, next time I'd like to see something a bit more realistic, like HT support disabled in BIOS, host and guest limited to say 4 CPUs (host via kernel parameters).

I would second that. You simply gave the host no time to take a breath.

You mentioned that the tests were automated. It would be very nice to see the test results in a more realistic environment, including vmware and openvz (if possible).

[latebeat]

Too bad it would be "illegal" to include any Vmware results. Oh well...
I wouldn't mind seeing some Xen comparisons though.
Great article anyhow

[deanjo]

Too bad it would be "illegal" to include any Vmware results. Oh well...
I wouldn't mind seeing some Xen comparisons though.
Great article anyhow

Well that's not entirely true. You may benchmark vmware and publish the results, you just have to send them your test plan, they approve the methodology, you benchmark, send results in and they either approve or disapprove to the publication.

[latebeat]

Well that's not entirely true. You may benchmark vmware and publish the results, you just have to send them your test plan, they approve the methodology, you benchmark, send results in and they either approve or disapprove to the publication.

Yeah right..
To me that's ridiculous... also judging from the amount of good and unbiased vmware comparisons that you can find online, you can imagine how often they "approve" the results :-)