As to your statement that VMware vSphere is capable of running more VMs per host than Hyper-V, what is that based on? Certainly the vendors specifications are covered in the tables in the whitepaper, and in fact according to the specifications Hyper-V can have 512 guests while vSphere can only have 256 per host. But practically, it is likely that because vSphere has more advanced memory management, it can run more guests than Hyper-V could on a given server. I know of no proof of that assertion however. If you have data, please send it along. Because that is an important and seemingly open issue, we are planning on testing memory utilization of the two virtualization platforms in our labs and posting our results here. Feel free to subscribe to our RSS feed or check back for updates.

–Peter

Marc.

In the white paper we discuss both the low-level hypervisors (like ESX and Hyper-V) as well as the virtual machine manager eco-systems like vSphere and Microsoft Systems Center. You are certainly correct that ESX (and ESXi) are free (but you have to pay for maintenance). Just like Hyper-V is free with Windows Server 2008 R2. But no one (or at least very few sites) run just the hypervisor. You need the management infrastructure as well, which is why we include that in the discussion. Also in the white paper is a cost analysis of a couple of scenarios, including hypervisors, management, and maintenance.

Technically there is certainly a debate about whether Hyper-V is a type 1 or type 2 hypervisor. That’s also discussed in the white paper. But my take is that Hyper-V is closer to a type 1 than a type 2. If it’s stable, functional, and well-performing, then it qualifies for production use (in my book). There is no thin deployment of Hyper-V though, unlike ESXi, so I feel VMware has an advantage there.

–Peter