Well, I'd start by looking at the business case. When your initiative was approved there should have been a business...
case associated with it that identified certain objectives that virtualization was designed to achieve in your organization. This probably included lowering hardware costs by increasing utilization, but it might also have included improving reliability, shortening the server provisioning cycle, etc. All of those things have value that could figure into your ROI calculation, but none of them are simple to determine.
Since the compute capacity of your server farm doesn't stand still, you have to find a way of tracking how much compute capacity is utilized at any particular time. Virtualization initiatives should drive your hardware cost per unit of utilized compute capacity downward. Compass uses hardware cost per utilized CSpec for this key performance indicator. If your cost drops by, say, $1 per month per utilized CSpec, all you have to do is multiply that by the total number of utilized CSpec in the farm to get the total savings. CSpec is proprietary, but other publicly available measures such as SpecInt can also be used.
Measuring the value of downtime avoidance involves risk analysis that your firm may or may not have done for the server applications you are concerned with. If they have done it then you should have an estimate of what a minute of downtime costs. Unfortunately your reliability benefits are going to take time to prove themselves. You need at least 6 months for a good sample of downtime numbers from the newly configured server environment.
In some companies provisioning time can be everything – the financial benefit associated with getting new applications set up on new servers faster can be in the millions of dollars when critical new applications are involved. For other companies provisioning time is much less critical. If you fall into the former category, you'll need to look at what applications have been provisioned in the new environment and whether they were up and running faster than they would have been pre-virtualization. Then you'll need to talk to the business units to get them to help you put a value on that time to market. Again, you'll probably need at least a 6 month window to collect meaningful data on this.
Related Q&A from Scott Feuless
Expert Scott Feuless answers a user question on which types of virtualization technologies he takes into account when speaking of virtualization.continue reading
Does implementing a virtual architecture mean IT departments will need fewer workers, or will current IT workers simply need more training? Scott ...continue reading
Learn why virtual sprawl is a problem in virtual environments and how it happens in this expert response from Scott Feuless.continue reading
Have a question for an expert?
Please add a title for your question
Get answers from a TechTarget expert on whatever's puzzling you.