My organisation went down the Nutanix path with about 1/4 of the DC about 18 months ago. They're now just finalising the move away feom Nutanix. From a server and dev admin point of view we had really odd VM behaviour, poorer than expected process performance, and random instability that just couldn't be resolved. I believe other system owners had similar and that the VM admins had their own range of oddities to track down. Something behind the scenes was the catalyst for change away in a short period of time.