In order to do virtualisation properly and safely without affecting the response time of the system during peak loads. We must have historic performance data for systems we are virtualising. At least 18 months, to cater for seasonal usage patterns. If not we may not be able to meet the Service Level Agreements (SLA's) due to bad response time and in the worst case a system could crash if overloaded.

System crash is not the bad thing, it is the affect this has on the people who depend on the computer system that cannot do what they wanted to.

So for people doing virtualisation correctly, they will need to monitor, analyse trends and generally get better at capacity planning.

I expect a silent boom in IT capacity planning.