Q&A: Virtualization -- From ROI to Optimization

From determining the ROI to ensuring you're getting the most from the technology, an infrastructure architect offers tips for IT professionals.

Whenever IT undertakes a new project, a project manager must be prepared to answer a key question: what's the value of this project to the organization? IT must overcome any objections, and one effective way to handle this is to present upper management with a thorough ROI analysis. Determining the ROI of a project -- especially one that involves virtualization -- can be tricky.

After a project is approved, the real work begins -- but it's only the beginning. IT must always ask itself how it can get the most out of a technology.

We explore these ideas with Steve Fink, a senior infrastructure architect in the Americas CTO Organization at Avanade, a global IT consultancy. Steve is responsible for managing the opportunity pipeline with Accenture's Technology Consulting, Data Center Technologies, and Operations organization. Steve also plays a role in defining Avanade's "Next Generation Data Center" strategy, particularly concerning physical data center design and virtualization.

Enterprise Strategies: What are the factors that people often overlook or misunderstand when they are trying to figure out their virtualization ROI?

Steve Fink: People look at components of virtualization in terms or ROI, but it stretches far beyond the number of instances that you can put on a host hardware device. The TCO extends into areas such as power, cooling, data center floor space allocation -- even carbon footprint for some companies.

What are the common misperceptions around virtualization ROI?

One common misperception is that no matter what you do to virtualize, you still have X number of OS images. That's true, but a virtualization exercise permits the operations and engineering groups to take a fresh look at a common OS image. After a common OS server image exists, it is far easier to use automated tools to manage administratively, such as patch management, security management, enterprise management -- even reboots.

What are you see from customers in terms of savings, costs, etc.?

We are seeing server consolidation ratios as high as 18-19:1. With that comes a plethora of savings -- power costs are reduced at a several orders of magnitude, IT can realize raised floor space savings, or alleviate an impending floor space crunch. HVAC costs are significantly reduced. Even more so, we are seeing an efficiency multiplier of approximately three through management tools, common OS platforms, revisiting how applications are supported.

When departments begin to cost-justify a project, they often have to contend with trying to sell a project on its "cost avoidance" benefits, but those are frequently difficult to quantify. It sounds like the cost benefits of virtualization are much easier to calculate.

Cost avoidance can be thought about in multiple levels. I don't think the 80/20 rule really applies here. Some of the obvious costs -- such as not purchasing unvirtualized hardware in a tech refresh and purchasing virtualized platforms instead, are easy to calculate. However, it becomes a less obvious, more challenging study on other categories, for instance, FTE savings. We can do it, but that's going to be on a case-by-case basis, and what works for company "A" will not necessarily be applicable for company "B." Even so, be prepared for a large pushback, as various decision makers and/or influencers at a company may have varying perspectives and differing thought.

What kinds of companies are doing virtualization (i.e., global companies, etc.)?

Nearly every company has been in the virtualization area for several years now. Keep in mind, virtualization really extends back to the 1960s in the mainframe world with LPARs. With VMWare bringing their product line, and Microsoft introducing Hyper-V, virtualization has been in development, test and QA environments for several years at most Fortune 1000 companies. As the two leading vendors for virtualization continue to deliver proven stability and their products continue to mature, this same corporate universe is warming to the idea of extending virtualization concepts to production.

What are the biggest impediments to adopting -- or getting the most out of -- virtualization?

Process, process, process. Any organization that is giving serious consideration to, or about to adopt, virtualization has to have their operational process house in very good order. With virtualization, another layer of operational consideration is being implemented -- the host hardware and software -- which has a far larger reaching implication if it goes out of service.

The mindset needs to be adopted from "delivering an application on a server" to "delivering a service through x, y, and z". Delivering that service has far greater reaching considerations to evaluate when making changes.

What approaches has IT been taking to overcoming these problems?

I think that the ITIL framework is a good baseline start. ITIL, though, is not for everyone, and no implementation, where it does make sense, will mirror another implementation. Each company has its own priorities, which need to be matched to process capability.

That being said, to move to server virtualization, as I mentioned earlier, companies must have their operational processes solidly in order before moving to the technical piece. Without solid operational processes in order, the technical side of the server virtualization effort will likely fail. This will include configuration management, change management and, release planning, and management.

We've seen virtualized servers and storage virtualization is getting a lot of buzz lately. Where is virtualization heading?

At Avanade, we believe that server virtualization will continue to proliferate within the operating environments. The take up has been rather strong in DEV, TST, & QUAL server environments for several years already. And, with the introduction of Microsoft Windows 2008, the Hyper-V virtualization suite, and evolution of Microsoft System Center suites: Operations Manager, Virtual Machine Manager, and Configuration Manager, it will go even further.

These products will help to mainstream virtualization technologies into customers that have been resistant to explore virtualization in the past as well as provide additional options at attractive price points for existing virtualization customers. This will also really escalate the take-up in the PRD environment sector.