by Debby Young

Changing the game plan from cutting costs to increasing service.

With server virtualization, IT can dynamically provision resources for the corporate computing environment based on anticipated workload cycles, such as normal end-of-the-month spikes caused by payroll processing or increased demand on online ordering resources triggered by a planned promotion. Because virtualization can emulate multiple computer environments on any given server, IT can pool server resources across the enterprise, thus driving down the cost of operation. This just-in-time resource allocation is part of a predictive resource scheduling strategy that optimizes utilization and assures service levels despite fluctuating workloads.

"Ultimately, you want to be able to allocate resources based on business priorities," says David Cohen, vice president, research and development for Merrill Lynch. He likens it to the way the electric company distinguishes between hospitals and residences -- during a power outage, hospitals get priority service over the general public. "In a virtualized environment, IT can configure resources to enable mission-critical applications to take precedence over less vital applications when extra processing power is needed," he says.

Virtualizing such tasks as data storage and network support can reap similar efficiencies. Depending on its failover configuration, for instance, pooling standby servers alone could save a company upwards of 40 percent on idle infrastructure expenses, floor space, power consumption, and support personnel.

In the multitiered application environments endemic to large enterprises, IT tends to overprovision application clusters because need is calculated on the peak requirements of every application. By virtualizing servers, resources within the server cluster can be continually repurposed to meet changing capacity needs. When demand subsides from one application, the resources can be redirected to another application experiencing peak load. Therefore, the cluster can be provisioned closer to average requirements rather than to peak loads, optimizing resource usage and lowering the cost of ownership without compromising service-level agreements. In baseball terms, it's like having a utility player available instead of a specialized backup player for every position.

Because virtualization separates applications from the platforms they run on, disaster recovery -- one of today's key business priorities -- is quicker than in traditional IT environments. For instance, with a physical server, if hardware is modified after software is installed, the data restoration might fail because the licensing key no longer recognizes the configuration. "In a virtualized environment, from the operating system's perspective, it's all the same hardware," explains David Boyes, president and chief technologist for the Ashburn, Va.-based R&D company Sine Nomine Associates. "This can take literally hours and days off your disaster recovery time."

The Latest Advances

Advances in virtualization technology are occurring on a number of fronts. Foremost is dynamic, orchestrated provisioning -- that is, quickly reallocating servers from one pool of applications to another. This sophisticated process involves reconfiguring server parameters, allocating storage and other resources on the fly to meet the increased work-load demands of another application. VMotion technology from VMware, for example, allows IT architects and administrators to view the server farm as one aggregate computing pool and carve off logical servers to meet peak loads or to test new applications with no perceptible delay in service.

VMware is also innovating how memory is managed in virtualized, consolidated environments. With advanced memory management (AMM), IT no longer needs to calculate memory requirements based on the total amount of memory in each of the systems being consolidated. "Because AMM optimizes how the overall system is used, IT can often reduce actual memory requirements by 50 percent or more," explains Brian Byun, VP of software alliances for VMware.

In addition, progress is being made in balancing workloads across clusters in the enterprise to meet the service-level agreements for business applications. The goal is to prevent the reallocation of resources without first determining the underlying relevance to the business. For instance, is the spike in transactions in the Web store caused by an increase in purchases, or is it a denial-of-service attack? If it is the former, IT would need to reallocate resources to keep up response times in order to avoid losing sales. If it is the latter, IT would have to throttle back resources and address the attack.

Virtualization technology is rapidly gaining adoption. Innovators such as Merrill Lynch are using it to ease the transition to new computing platforms and manage application updates. For instance, when Merrill Lynch rolled out Windows XP across its user base, VMware allowed the company's financial advisors to toggle between Windows NT and Windows 2000 on the same desktop so that there was no disruption in service.

Industry experts observe that virtualization will enhance IT's ability to seamlessly integrate with partners that can provide additional resource capacity to handle workload spikes. There may be a slight premium on those resources, but it will cost a company far less than it would to retool the enterprise infrastructure to support peak requirements.

--

Additional Reading - Sponsor Link:
Seven Requirements for Balancing Control and Agility in the Virtual Environment