While this is a good question, and one that certainly should be asked by IT managers, it really is best to proceed with caution and not depend on the projected numbers or industry averages. Many great organizations loose valuable people in their aggressive push to meet such numbers in the fantasy world of virtualization. Attrition and internal reassignments are the best way to deal with the operational advantages afforded through consolidation.
With that said, the average admin:server ratio is typically between 1:10 to 1:30 for non-consolidated environments. Although many pre-consolidation figures are way higher (i.e. 1:100 or 1:1000). This is either because the IT staff is already drastically overworked or in google's case through extreme standardization (tens of thousands of identical servers).
See page 15 of the following VMware whitepaper that discusses the ratios based on application type and support level:
Improving_ROI[/url].
Notice I didn't give any post consol numbers, because I think that will vary greatly. I've seen many companies lay off 60% of their staff, just to hire more back at a premium a year later.
However, Gartner states that 70% of virtualization savings come from staff reductions, and IDC's John Humphrey boasts[/url] admin:server ratio increases from 1:20 (pre-consol) to 1:200 (post consol). While such numbers very well can and do happen, it's not something to bank on long term.
Just proceed with caution as there is no magic formula for determining the numbers you will get, or the intangible losses you may face through loss of knowledge capital, and customer satisfaction while staffing issues are ironed out.
Reality is though, usually after a tough year for everyone, the company has saved millions, the dead weight is gone, new opportunities have opened up for motivated internal staff, and the company rebuilds to accomplish great things.
Just one man's perspective on a not so simple question :smileysilly:
Hopefully others will chime in with what you "really" wanted to hear, some hard 'before' and 'after' numbers. Optional, but helpful data would include what percentage are (DEV, QA, and Prod) divided by the (# of VMadmins).
cheers and best of luck.