You virtualized your first production server.
And it was good. It was great. You repurposed some VMware licenses from your test lab, grabbed some under-utilized hardware and went to town. Print servers? Check. Ten to one consolidation. Secondary domain controllers? Why not put one on the print host? Great. Marketing database running at single-digit CPU loads? Why not guest that on the new eight-way box in the lab; Marketing will never know...
Another year rolls around, your ops budget received formal approval for an ESX rollout and you learned the joys of vMotioning across 20 hosts. Downtime is a thing of the past. Your x86 service manager is queued up for a fat performance bonus; delivery and performance metrics for service have never been better.
Then it happens. The hairs on the back of your neck start to itch.