Determining what kind of return-on-investment (ROI) you can get out of a virtualization project – and who's going...
to pay for it -- is the new budgeting puzzle facing IT managers, says IT finance consultant John Hayes of Avnet Technology Solutions.
Unfortunately, there are not a lot of tools for handling these tasks; but in his field work, Hayes has helped IT organizations make do with a few, including VMware Capacity Planner.
In this interview, Hayes describes the challenges inherent in budgeting and planning virtualization projects and gives some tool tips. SearchServerVirtualization.com caught up with Hayes just before today's tutorial, "Determining ROI for Consolidating with Blades and Virtualization" at Server Blade Summit in Anaheim, Calif.
SearchServerVirtualization.com: Has virtualization thrown a monkey wrench into old-school IT accounting practices?
John Hayes: One would think that such a large shift would have a ripple effect; but, no, accounting practices are based on more long-standing rules and traditions.
Virtualization certainly creates challenges for companies that have implemented chargeback mechanisms in their P&Ls. When physical resources are shared, it becomes very difficult to determine what to charge who. The assessed costs would be very dynamic.
For example, if department A needs a development server and creates another VM (virtual machine) on an existing server that department A is sharing with department B, should department B's chargeback be lowered by some amount?
Are there other cost management problems that occur when virtualizing?
Hayes: Generally the company will face an equipment disposal event. We tend to obsolete about 90% of their equipment when we virtualize their environment. So, they have to figure out what equipment can be redeployed and what should be just written off and disposed of.
Is it hard for companies to part with that existing equipment?
Hayes: Some get way too hung up in the investment they've made in their existing equipment, thinking if their servers are newer they aren't candidates for virtualization or blades. They're losing sight of the real cost of operation of the environment. It's not the hardware. It's the maintenance, administration, power, refresh cost and data center infrastructure costs.
Are there other financial concerns that IT managers grapple with when virtualizing?
Hayes: Companies way overestimate the costs of implementing a virtualized environment. It's not that hard. And, some struggle with how to come up with the funding to bring in all this new beneficial technology. They can't get their head around the fact that it pays for itself in a real short timeframe out of money they are already spending.
What are the ROI results you've seen when companies virtualized on blades rather than racks?
Hayes: Under certain conditions, blades can be less expensive than the equivalent rack solution, and hence there is a higher ROI. Generally this would be for a situation where you're dealing with more than 10 servers in a single location and the client's workloads are memory bound (not disk or CPU). Luckily, about 50% of the Windows Server consolidations we've performed fit this situation quite nicely. I'm presenting an example during my session showing a 40% difference between implementing a blade solution vs. rack solution.
What are some methodologies for forecasting ROI when consolidating servers via virtualization?
Hayes: We use the methodology outlined by Brigham and Gapenski in "Intermediate Financial Management" (The Dryden Press).
If you're an IT manager, it's definitely a benefit to use a tool that your finance department is familiar and comfortable with. It allows you to represent your projects in a language they can directly relate to.
From your experience, what are the most-used financial and capacity planning tools for virtualized environments?
Hayes: VMware Capacity Planner (formerly AOG) and CDAT have emerged as the two leading tools. This has been a three-to-four year evolution, but they've been the most popular choices for at least a couple of years.
We use Capacity Planner for the capacity planning part of the job; i.e., analyzing the client's current environment and building consolidation scenarios. We particularly like it because it is agentless, and it does a great job of figuring out the primary consolidation constraints; that is, what is it about the client's actual workloads that are the most important factor to consider when picking a new target server environment.
For the financial part, the Brigham Gapenski framework is very powerful.
Could you offer some tips for using VMware Capacity Planner?
Hayes: Like CDAT, Capacity Planner is great for showing you what is possible. But it doesn't show you what is practical or financially viable or optimum. To get that you need to couple the technical tool with a financial analysis, and neither CDAT nor Capacity Planner do a rigorous job of that.
Capacity Planner is a very complex tool. It's like any tool, in the hands of an inexperienced craftsman you're not going to get a good result.