So you want to jump on this virtualization bandwagon?  You too want to tout your “green” initiatives so you can be the envy of all those others gathered around the 80% post consumer recycled water cooler.  How on earth can you get the management buy in that’s needed?  Impress the boss with large statistic graphs on lower overhead cost, lower power consumption, and a smaller computing footprint?  How about your lab it?

That right, why don’t you lab it.  Take and model your application servers, model your test environment, model the interconnections, and prove that virtualization won’t impact the business in a negative way.  What I’ve found is that almost anything can run in a virtual machine.  I’ve also found that a number of applications don’t run well in a virtual machine.  As infrastructures and VM technologies improve I think this is going to change, but…  Until is a sure fire thing, why not test it?  The things to consider:

What will the new virtualization platform look like?

• Will it be based off of blades?
• Do I need a SAN?
• How much hardware will have to be bought?
• How can I liquidate the hardware that I won’t be using?

What are the most critical applications that need to be modeled?

• Which one will I lose my job over if it goes wrong?
• Which one keeps me up at night because I’m nursing something in dire need of a technical refresh?
• Which one is the easiest to rebuild?

What are my success criteria?

• Is I/O throughput important?
• User experience?

Taking and modeling, testing, poking and prodding your environment in this manner allows an opportunity to reduce the impact of the learning curve when your shiny new VM environment is in production.  This approach allows you to see what your users will experience in order to manage their expectations and fears.  It also allows you to have one great way to prove the benefits of a VM approach versus the single server single purpose mindset.