The two things that every CIO should know before going virtual are surprisingly simple. First, the biggest benefits are not the ones that your infrastructure folks are using to try to convince you to take the plunge. The biggest benefits are the ones from which application owners-the folks who are probably pushing back on virtualization-are most likely to benefit. And second, if you get it wrong, the risks are large enough to wipe out all the benefits you hope to realize from virtualization. Fortunately, these risks are completely avoidable.
Infrastructure folks will tend to focus on utilization and that makes a lot of sense. The utilization of an average, nonvirtualized server is something south of 20 percent. By consolidating the work loads of underutilized servers, many CIOs should be able to reduce the number of physical servers running these work loads by a factor of four, conservatively. Fewer servers mean less power, cooling and floor space consumption.
Application owners will often push back, concerned about the overhead of virtualization and sharing resources with other applications. Besides, they correctly point out that licensing and support costs will wipe out much of the savings associated with fielding fewer servers.
Two reasons to go virtual
What CIOs need to know is that virtualization will reduce the number of servers and increase their licensing and support costs. But, rather than focusing on cost saving, CIOs should focus on the top line.
The principal reason to go virtual is to increase IT’s agility-the ability to react to changing market conditions. By doing so, CIOs will help their organizations convert shorter time to market (TTM) into market share and top-line revenue growth. Ironically, application owners are the ones most likely to benefit from greater agility, while the infrastructure folks will be forced to combat virtual server sprawl.
The second thing CIOs need to know about virtualization is that it is easy for the benefits to be overwhelmed by downside risks. Virtual environments are more complex and difficult for system administrators to visualize. Problems can take longer to resolve unless the proper tools are in place.
Fortunately, the downside risks associated with virtualization are avoidable. The key is effective application performance management (APM). Effective APM enables early detection and quick resolution of problems, even in complex virtual environments.
Since even the most aggressive enterprises are expected to eventually virtualize as much as 80 percent of their infrastructure (the average today is approximately 20 percent), most applications will span virtualized and nonvirtualized infrastructures. This means that effective APM tools must support end-to-end visibility across the enterprise’s entire infrastructure portfolio-including mainframes in many cases.
There are APM tools available today that can bridge the gap between virtual and physical infrastructures, as well as reduce the complexity associated with data center virtualization. These APM tools do this by providing a unified view of the entire data center where physical and virtual infrastructures are visible together in a single view.
Doug Willoughby is Director of Cloud Computing at Compuware Corporation. Before Compuware, he was at Sun. Doug joined Sun in 1988 and participated in the development and marketing of some of Sun’s most pioneering technologies including Project Spring, Distribute Objects Everywhere (DOE), NextStep/OpenStep, and Java. Doug was also part of the team of 14 engineers and architects who developed “network.com,” Sun’s first utility computing offering. Doug has published numerous articles and speaks regularly at various industry events. He can be reached at [email protected].