Hyper-V Server + Windows cluster + SCVMM in most workload scenarios is now "good enough" compared to other solutions, and is usually (not always, but usually) quite a bit cheaper to run.
I concur. As of Server 2008 R2 SP1, and SCVMM 2008 R2, Microsoft’s solutions became good enough compared to VMware’s solution. As of the 2012 product line, they really became equal's. As of 2012 SP1, Microsoft's 2012 virtualization products appear to 100% match or exceed vSphere. There may be specialized use cases that VMware solutions are still better for even compared to SC2012 SP1, but I don't know of any right now.
I work in education. Our servers were licensed for vSphere as the hypervisor, vCenter to manage vSphere, Windows Server and Linux as the guest VM’s, which we actually managed with System Center. This was the correct decision in 2004. It was the wrong decision as of 2011. 2008 R2 SP1 gained Dynamic Memory, the System Center 2012 CTP came out and proved Microsoft was going to be able to completely replace vCenter, and SCVMM 2008 R2 was good enough in the midterm. We took 1/5th
of our production environment, and moved it to Hyper-V and SCVMM so we would be prepared to take greater action if System Center 2012 was all it should have been at release, which it mostly was. All the flaws in SC 2012 I’ve seen are resolved in the SP1 beta, including chargeback.
At this point we’re about 70% Hyper-V and 30% vSphere. They are fully comparable products. In fact, I would say that multi-processor support "feels" better in Hyper-V 3.0 than even in ESXi 5.1. I personally would like to see third-party benchmarks comparing something like a large SharePoint environment
using 16+ virtual processors compared against the two products. In vSphere 5.0, the more processors you use, the worse it performs due to its processor queuing system (as I understand it anyway, ESXi 5.1 does it better, but I don't really know how much better). I haven't done direct benchmarks, I can't give quantifiable
data, but Hyper-V "appears" to offer VM's greater overall I/O and performance than what I've seen in vSphere. I've barely used ESXi 5.1, and i'm not likely to progress further on that platform, so I'm not likely to generate much more of an opinion comparing
the two in terms of raw VM performance.