LAS VEGAS -- Too many midmarket CIOs look at server virtualization as a means of reducing space, power and cooling...
By submitting your email address, you agree to receive emails regarding relevant topic offers from TechTarget and its partners. You can withdraw your consent at any time. Contact TechTarget at 275 Grove Street, Newton, MA.
demands in the data center. As a result, they could be shortchanging their businesses.
Realizing savings from consolidation is largely a onetime event, said Gartner Inc. vice president Tom Bittman, who spoke Tuesday at the Gartner Data Center Conference.
In fact, if organizations are getting into virtualization just to address power and cooling issues, they might be making a mistake.
"You need to be proactive about planning," Bittman said. "If you're running to virtualization just to save money on space and power, you might be going in the wrong direction. Virtualization gives you a lot of capabilities to deliver, but the idea of solving server sprawl with virtual machine sprawl is the wrong idea. You have to have very good management capabilities to keep that under control."
IT organizations should take a more strategic view of virtualized infrastructure, he said. Virtualization allows for faster provisioning and deployment because it's easier and faster to deploy an application on a virtual machine than it is to deploy it on a newly purchased server.
"Deployment time moves from days or hours to weeks or months," Bittman said. "It usually takes two or three months to get requests from the business to get a server ordered, wired up, deployed and the application installed and ready to go. With virtual servers we see that changing to two or three days."
Along with planned downtime, the migration of virtual machines from one server to another also makes disaster recovery more affordable for organizations. Virtual machines running critical applications can be migrated to off-site servers during a disaster. Bittman said many organizations that are deploying virtualization for the first time are also trying disaster recovery for the first time.
"From a planning perspective, we typically plan things in silos," Bittman said. "We're going to move to a more holistic model. We're going to plan things over a range of servers, storage and networking."
The days of hardware spending being project-based will become a thing of the past, Bittman said. No longer will IT buy a new server to run a new application. Instead, infrastructure spending and application projects will become asynchronous.
However, virtualization has a dark side, according to Bittman. Virtualization increases service demands from the business, so IT needs to be ready. Virtualization eases the "friction," or barriers to entry, of getting new capabilities from IT. When business units perceive that they can get new capabilities easily, they will start making more demands.
"When you take friction out of the equation, companies ask for more. Companies going to rapidly virtualized environments have seen demand skyrocket."
Bittman said IT organizations need to prepare a chargeback model that tracks usage of the virtualized infrastructure by business units.
"There are some really good things about being able to deliver what the business needs, but the business should pay for that. You need a chargeback model."
David Heiser, desktop and server services manager at The University of Nevada, Las Vegas, is just starting to evaluate virtualization. Server consolidation remains top of mind for his virtualization plans. However, he said he also hopes to leverage the technology for disaster recovery.
Heiser said the 150 servers at his university have no more than a 10% utilization rate, but his infrastructure is at the absolute limit in terms of space, power and cooling capacity. To deliver more services, he needs to virtualize his servers.
Heiser said that through consolidation, he not only anticipates being able to add more applications, but he also believes he will be able to concentrate the university's 25 most critical applications on one or two physical machines. In a disaster recovery scenario, he'll be able to keep those applications running longer on UPS power by running fewer servers to support them.
Heiser also has his eye on desktop virtualization.
"The next step is going to virtualization of desktops so we can push and control what's going on with machines and know what customers are running on them. Most problems you have with PC operating systems, everyone has administrative rights. Problems are induced by users. We have to go do repairs that are created by customer errors rather than system errors. Desktop virtualization would eliminate that."
Eric Brown, associate director of IT at Gilead Sciences Inc., a 2,700-person biopharmaceutical company based in Foster City, Calif., said he's "riding virtualization pretty hard."
Brown said his company is running 50% of its applications on virtualized servers. Gilead's goal is to virtualize no less than 90% of its servers and nearly all of its storage.
Gilead jumped headfirst into virtualization for many of the reasons Bittman cited. Brown said reducing the server infrastructure's energy and physical footprints were a priority, but there were also some other benefits Gilead wanted to realize, such as zero-impact maintenance, better disaster recovery capabilities, quicker application provisioning and deployment.
Gilead is capable of establishing a chargeback model for virtual server usage, but his company isn't interested in it from an accounting point of view. Instead, such a model would be used to help his organization understand consumption of services.
Brown said the success of server virtualization at Gilead has increased demand from the business, as Bittman warned. He said his organization deals with this by educating business units.
Let us know what you think about the story; email Shamus McGillicuddy, News Writer.