VMware Outperforms Hyper-V
All of you have probably heard of the VMware Cost-Per-Application Calculator which was launched a few weeks ago. To prove that VMware has a lower cost per application and a higher virtual machine density, The Taneja Group conducted a series of tests showing VMware’s free ESXi outperforming Microsoft’s Hyper-V in each of these areas. Feel free to read the entire document but for those who are time constrained, here is a summary of their technical findings:
- Hyper-V is still an immature product with limitations such as memory overcommitment and limited SMP vCPU support for older Windows OSs and Linux. Did you know Hyper-V only supports one Linux flavor with only one virtual processor? Anyone actually running Hyper-V Linux guests in production?
- Using a DBHammer workload in the guest, you can run up to 100% more VMs on ESXi than on Hyper-V. Actually, there was no technical limitation in creating even more VMs under ESXi, unlike on Hyper-V:
- On a CPU intensive workload (SPECjbb), ESXi delivers 24% more performance than Hyper-V running on the same hardware. This is mainly due to ESXi leveraging large pages and RVI. Hyper-V does not support these technologies.
- So what does this all mean to you? Utilizing VMware’s technology, VMware can be 5% to 29% less expensive than Microsoft.
We encourage everyone to try the calculator to see how much they can save with VMware. Here is a quick video demo on how to use the tool, presented by the creator himself, Alberto Farronato. You’ll find his accent quite sexy:

Microsoft addresses a couple of the VMWare myths here: http://blogs.technet.com/virtualization/archive/2009/04/03/Top-10-VMWare-myths-video.aspx
Patrick
April 3, 2009 at 9:53 am
Is that Patrick O’Rourke from Microsoft? Nice to see you comment on our blog! Thanks for stopping by.
Tim Stephan
April 3, 2009 at 12:39 pm
Thanks for the link Patrick. We’ll let the user comments to the video speak for themselves: http://tr.im/idLt
Michael Hong
April 3, 2009 at 1:55 pm
Apparently the Microsoft myth busting exercise busted itself – http://www.gabesvirtualworld.com/?p=445
Alberto Farronato
April 6, 2009 at 10:00 am
Leveraging RVI, eh? So you can only get acceptable performance on CPUs that have been shipping for the last twenty minutes?
David Shaw
April 29, 2009 at 11:40 pm