If this is indeed accurate, then this is big news. The Hypervisor recently published a comparison between VMware Workstation 6 and Hyper-V, and found that VMware Workstation 6 was faster than Hyper-V for a variety of tasks.
This is quite surprising, and I’m hoping we won’t find out that the testers installed VMware Tools for VMware Workstation but didn’t use the enlightenments for Hyper-V (or some other similar move). That would just be wrong. If you’re going to use optimizations available on one platform, you should also use optimizations on the other platform as well.
In the meantime, if anyone turns up more information on this comparison and the details behind the comparison, I’d love to hear it.
Tags: HyperV, Microsoft, Virtualization, VMware, Windows
6 comments
Comments feed for this article
Trackback link
http://blog.scottlowe.org/2008/08/01/vmware-workstation-beats-hyper-v/trackback/
Friday, August 1, 2008 at 9:54 pm
David Marshall
Hey Scott, great points there! I’d also like to see these folks run the same battery of tests against ESX if they haven’t. I’m curious to see what the results would be here as well. One reason being, back in the early days, I used to run quite a few comparisons between Workstation, GSX Server, Connectix and Microsoft Virtual Server and ESX. And in many cases, Workstation, GSX Server and Virtual Server would perform better than ESX.
This was only true when it was a single live virtual machine test per platform. Once you increased the number of live virtual machines, ESX would perform better than GSX and Virtual Server… those platforms would suffer in performance as it scaled.
So if this was a single virtual machine instance, can we get them to run the same tests again under ESX and also find the answers to your above questions from the original test?
Great find. Thanks for sharing it.
Friday, August 1, 2008 at 9:57 pm
slowe
It’s a sensational headline, no doubt, but we really do need to see the testing methodology. Once we can take a look at the methodology and procedures used in this test, then we can begin to determine the applicability of this particular benchmark.
Thanks for reading!
Saturday, August 2, 2008 at 4:58 am
Stu Fox
It’s a meaningless test. Installation is slow on Hyper-V, as during installation you can’t take advantage of all the speed enhancements the integration components give you. As soon as you’ve got the IC’s installed that’s when performance really kicks in, until then you’re using emulated hardware, and suffering a performance hit.
Disclaimer: I work for Microsoft NZ.
Saturday, August 2, 2008 at 10:02 am
slowe
Stu,
I appreciate you making your affiliation known.
I think everyone hear agrees that any test can be made to say just about anything, depending upon how you structure the test. That’s why everyone here appears to want to know the details behind the test, so that we can see just how valid the test really is (or isn’t).
Of course, something to consider here is that during installation VMware’s optimized drivers won’t be installed either, so both platforms are using emulation. This would imply that VMware’s emulation engine is more efficient than Microsoft’s.
Thanks for reading!
Saturday, August 2, 2008 at 5:14 pm
williamwbishop
How is this a surprise to anyone? Is this even newsworthy?
Hyper V is in first release, it will take a couple revisions to become even a decent technical platform(and it’s microsoft so, it may take a couple extra).
All the testing I’ve done, and seen done, has shown ESX to be the superior platform…But it’s at version 3.5 and growing.
Saturday, August 2, 2008 at 10:38 pm
Stu Fox
It’s not the way the test is structured as such, it’s a known case where Hyper-V performance is slower than you would expect. Installation is known to be slow - and in a real world case you would (should?) be using sysprepped images to deploy, not installing. Like I said above, once you’re actually running and the IC’s are installed it’s a whole different story.