By Amy Larsen DeCarlo

CIOs searching for a more efficient and flexible way to harness processing power need look no further than virtualization technology. Virtualization, which essentially hides the characteristics of the underlying hardware from the operating systems and applications that run on them, can deliver the kind of resource efficiency, agility and cost savings that CIOs and their organizations need to stay competitive.

The numbers tell the story: Businesses aren't getting a big enough bang for their hardware buck. While organizations spend billions each year on servers, CPU utilization rates average a scant 15 percent, according to research firm Gartner. Over provisioning is the conventional workaround to this problem; companies react to demand spikes by adding more physical servers to accommodate peak demand. But with this approach, servers run underutilized most of the time. Not to mention that deploying a new server can be a long, drawn-out process.

Virtualization—long an established concept in the mainframe world—is now coming into its own in the distributed server environment, thanks to user demand for increasingly compelling virtualization solutions. Another advantage: applications vendors do not need to provide separate versions of their applications to run in virtualized environments. This is a core part of the value proposition: Organizations can run any off-the-shelf operating system or application in a virtual machine.

In fact, more than three-quarters of businesses that employ 500 or more staffers have either already deployed or are planning to deploy virtual servers, according to market watcher IDC. Analysts project these numbers will continue to rise. Forrester Research reports that of all companies that have already implemented a virtualized environment plan, 60 percent will add more virtual machines in the future. "Virtualization is going to revolutionize the way computing can be used in the enterprise," adds Theresa Lanowitz, founder and CEO of voke inc., an industry analysis firm.

While the enterprise has traditionally focused on operating environments and specific hardware types, virtualization changes all that by extracting the software's reliance on particular hardware, Lanowitz explains. Removing the application's dependence on the hardware operating environment opens up a whole range of applications, such as server consolidation and more efficient disaster recovery.

At its most basic level, virtualization helps businesses consolidate servers and achieve better utilization. By combining applications on fewer servers, CIOs and their organizations can reduce the amount of data center real estate, heat generation and power costs. "Putting multiple virtual servers on a single physical machine can simplify IT management," says Audrey Rasmussen, principal analyst with IT market analyst firm Ptak, Noel & Associates.

Healthy Growth
Companies move to virtualization for a long list of reasons. For WellSpan Health, a York, Pa., operator of two hospitals and numerous outpatient facilities, the move was spurred by growing pains. In 2005 WellSpan was running out of room in its data center. The space constraints were compounded because an application hardly ever uses just one server, says John Coleman, WellSpan's manager of technical services. Coleman says that a single application may actually interact with as many three, four or even more servers, including a Web server, backup server, a database server and the application server itself. This opened the door for unwanted and exponential hardware growth as WellSpan continued to deploy new applications.

While Coleman and his team wanted to keep expanding applications, they also wanted to limit the number of new physical servers they would install. To meet this dual-pronged goal, WellSpan began deploying virtual servers. Coleman started small, in early 2006, and then ramped up quickly. Currently, the health-care concern has roughly 75 virtual machines in place, running on eight physical servers.

WellSpan uses virtual servers as part of its refresh cycle and to install new applications. In the last six months, Coleman says, 80 percent of the company's new applications have been deployed on virtual servers. These systems run everything from active directory services to its pharmacy applications.

Virtual Continuity
But while server consolidation was the primary driver behind WellSpan's virtual server deployment, the organization has since realized even more important benefits from virtualization, such as greater continuity of operations and lower disaster recoveryrelated hardware costs. "The key is that hardware abstraction layer," Coleman says. Because virtual applications aren't tied directly to specific hardware, in the event of a hardware failure or a disaster, Coleman can replace a failed server with another without having to make sure that the hardware is identical.

What's more, by using virtual systems at a disaster recovery site approximately two miles from the data center, WellSpan can get away with fewer physical servers. Virtualization also benefits end users, who now suffer through much less downtime after a failure. The business itself also benefits, mainly by holding its costs down as a result of not needing to purchase so many underutilized servers.

Other companies find that virtualization can act as a foundation for data center automation. This, in turn, allows CIOs to reduce overall organizational risk and improve operational continuity. With data center automation, organizations should also find it easier to define consistent processes for reprovisioning, reconfiguring and reusing servers for different workloads and applications, says Donald LeClair, a CA senior vice president of R&D. "It's a lot easier to reprovision a machine in a virtualized environment than it is in a traditional one," he says. "It can also be much faster."

Why? Because while traditional servers can take weeks to deploy, a virtual server can be provisioned in just hours, even minutes. "If you are part of that retail environment, you need to handle double, triple, quadruple your normal customer load on big shopping days," says Andi Mann, a senior analyst with Enterprise Management Associates (EMA), an industry analysis and consulting firm. "Otherwise, people will just click away and go to another retailer."

To help, some software vendors are stepping up to the plate with more sophisticated virtualization capabilities. VMware of Palo Alto, Calif., for one, offers itsVMware Infrastructure 3 suite, which includes capabilities such as live migration of virtual machines from one physical server to another; dynamic realtime allocation of resources to virtual machines that need them most; and automatic restart of virtual machines affected by hardware failure. VMware's VMotion capability, for example, lets a CIO move a live virtual machine from one server to another without any end user interruption, explains Bogomil Balkansky, director of product marketing for VMware. "With this live migration capability," he adds, "companies can pretty much eliminate scheduled maintenance windows that are too disruptive to the business."

CA Unicenter® Advanced Systems Management(Unicenter ASM) removes some of the complexity associated with managing involved virtual environments, says Balkansky. The CA solution provides critical capabilities, including advanced platform discovery, that system administrators can use to aggregate performance and availability information. It also lets companies dynamically prioritize and allocate resources. "That way, data-center administrators can manage their physical as well as virtual environment using a familiar set of CA tools, instead of having to learn and use various user interfaces for physical/virtual/clustered systems," Balkansky says.

For many CIOs, managing virtual environments is a pressing need. Their organizations' demand for computing resources may vary dramatically from season to season, month to month, even day to day. Also, while the computing demands of some business processes rise and fall on schedule—say, once a month or once a quarter—the needs of many other processes are far less predictable. "Since computing-resource allocation needs to be done all the time, it should be automated," says Peter Richardson, a director of product management at CA. "Both CPU and memory may be needed, so the ability to allocate them and reclaim them independently and automatically is also important. Returning unneeded resources back to the pool is also important. Or, like inactive virtual machines, they can virtually disappear."

Business Connection
One measure of success for managing and optimizing a virtualized environment is connecting the business value of the application or IT service being hosted to the level of resources being assigned. In fact, aligning IT resources to business needs is one of the key concepts behind CA's EITM strategy. In a virtualized environment, this would be seen by defining policies and priorities for allocating resources to ensure that the IT services with the greatest business impact always have the capacity they need.

Yet there are also elements of considerable complexity associated with managing virtualization's inherently multilayered environment. "You not only have to be managing all the physical servers, but you also need to know what virtual systems are running on top of the physical hardware," explains Rasmussen of Ptak, Noel. "The virtual systems must be managed, too."

One reason for the complexity: There are a number of variations of the virtualization theme, and all of them can coexist in a single IT environment. In one version, multiple operating systems run on a single hardware host, creating virtual operating environments. In another form, multiple physical servers are clustered together to "look" like a single logical entity.

Also adding to the complexity, virtual environments are often populated by software and hardware from multiple vendors. Invariably, some applications are better suited to virtualized environments than others. This, in turn, can require a company to continue running conventional physical servers alongside its virtual environments.

For IT teams trying to create a cohesive and productive technology environment, having a single platform to manage the entire environment can be a key asset. While application requirements aren't going to change, the environment is evolving to include both shared resources and traditional physical servers. For this reason, having a system that can manage the entire environment means an organization can monitor the servers and ensure acceptable service levels, says Richardson with CA.

Luckily, many of the management capabilities needed to orchestrate a virtualized environment are similar to those needed to run a stable physical one. For example, businesses have to be able to discover the virtual environment as well as the physical servers virtualized applications run on. IT managers must also be able to configure the systems, which can range from simple virtualization to more complex, dynamic system resource management, Rasmussen says. While management applications are improving, all systems aren't created equal. "CIOs need to understand what aspects of virtualization management are being offered, because in some cases they may be doing just discovery and configuration, or a combination," Rasmussen says. "But other will also do software deployment and patch management, performance management and resource management."

One of the more cutting-edge applications for virtualization is automated workload management. It essentially takes job scheduling to the next level. While traditional job scheduling manages work by date and time, workload automation scheduling also manages by specific events or requirements in the application environment, explains Jim Anderson, director of product management with CA. In other words, workload automation gives companies a way to manage unplanned work.

Although cost often leads the discussion on virtualization, it's the desire to achieve acceptable service levels in high-transaction environments that's pushing many CIOs to investigate virtualization. Simply put, how can an organization manage and prioritize its work and provision its resources properly to handle that workload?

Too often, companies struggle to balance resources in a physical environment. If an unscheduled application requires more bandwidth than is available—perhaps because a batch process or a backup is being run at that time—the job may not be run at all. But in a virtualized environment, competition for resources can be resolved by a virtualization broker. This software finds the memory or processing power needed for the application to run.

At the moment, many companies are just beginning to implement dynamic workload automation, though VMware says about half its customers are migrating a live virtual machine from one server to another. Many businesses must still work through the basics before moving on to the more advanced applications for virtualization, says Rasmussen, adding, "The IT process discipline isn't there yet." By working on the basics, those companies will be able to employ dynamic workload automation later.

Virtualization could also be used for quality assurance and testing. Looking ahead, analyst Lanowitz predicts that software developers will create virtual production environments as a way to test their applications, rather than asking their environments in the lab. The advantages of this approach: Developers can save time because they don't have to set up a lab environment. The business can save money because it doesn't have to duplicate the production environment. And the company can get more accurate results, faster.

Amy Larsen DeCarlo is a Virginia-based freelance writer who has covered technology and business for more than a dozen years.