Consolidation and virtualization of physical servers is a substantial move toward increasing data center efficiency. Still, there are steps that can be taken to further maximize the virtualized architecture. These include:
To effectively deploy and manage virtualized systems, companies will likely need to cross-pollinate skill sets within the IT staff. Therefore, moving to virtualization can require some substantial investments in training.
The IT organization will have to work more closely than in the past. Furthermore, companies may also have to overcome possible turf issues as different IT managers and groups jockey for position.
People tend to be one of the biggest initial barriers to widespread virtualization, says John Humphreys, senior director of product marketing for Citrix Systems’ data center and cloud division. “Companies have taken virtualization and captured the low-hanging fruit,” he says. “Now, it’s a question of taking it from 25 to 30 percent of application workload to 60, 70 or 80 percent. That’s where a lot of the teaming and partnering and just general soft-selling comes in and plays a big role.”
Start by virtualizing a few general but not critical applications, Humphreys advises. It serves two purposes. First, these types of apps typically aren’t viewed as “owned” by specific users, departments or divisions.
Because many employees will likely use them, it will quickly expose staff to the virtual environment and help ease concerns. Users can see that the VM will support them just as their stand-alone server previously did.
A natural next step is to virtualize line-of-business applications. These often are “turf” apps, run or overseen by specific program managers in a company. Once the IT team has a few system “owners” on board, those users can influence employees who may be wary of virtualization.
Another focus should be managing the new environment. If unmanaged and adoption is widespread, uncontrolled or too rapid VM sprawl can result.
“Increased utilization comes with increased complexity,” says Paul Christman, vice president of state and local government and education sales at Quest Software. “A lot of folks found that they didn’t get the benefit they expected with virtual servers because complexity came along with the virtual infrastructure.
“You need new tools to help manage this explosion of virtual machines,” he says. VM sprawl refers to the point at which there are simply not enough people internally with the right skills or expertise to manage the growing virtualized environment.
Because it’s so easy to deploy virtual systems, IT can quickly run into management and capacity issues. “You can end up with rogue virtual machines consuming resources, with nobody actually using them,” says Andi Mann, vice president of virtualization product marketing for CA Technologies.
Having the right tools to manage virtual machines and the virtualization environment overall is essential. Such tools allow setting policies for virtual machine deployment, configuration, security and optimization.
In addition, policies should be determined for identifying and decommissioning VMs that may no longer be needed. Policies should also be set for VMs in operation, making sure they are running at optimal levels from a department resource standpoint.
Today, makers of server virtualization software — such as Citrix Systems, CA, Microsoft, Quest and VMware — offer a host of applications that allow the IT staff to readily monitor and manage virtual environments.
The next phase of server virtualization is not as obvious as the initial consolidation and migration, Humphreys points out. “It’s not just server consolidation anymore — it’s taking advantage of application mobility.”
Application mobility in a virtual environment simply refers to the ability to move an app seamlessly from one physical host to another. It might be done for any number of reasons.
For example, “suppose you want to cut down on the amount of power that you’re using at night. You have apps that are ‘at rest,’ so you roll them onto a single server at night while there’s hardly anyone using them, and turn off all those other servers that aren’t being utilized,” Humphreys explains.
Mobility also means a company can respond to spikes in demand in real time. “What you can do now is temporarily add capacity and then retire it as the spikes go up and down,” Christman says. “In the physical world, putting in load and taking it out intelligently based on real-world information is really hard. That’s one of the things virtualization provides.”
The SLA offers testament to the promise that applications and information processing systems, needed to successfully run the business, will be provided. Unfortunately, many companies do not have SLAs in place for their virtualization environments.
Anil Desai, an independent IT consultant in Austin, Texas, recommends creating detailed service level agreements to manage the process, monitor results and develop best practices. “When everything is virtual, you don’t have the comfort of seeing the classic blinking lights on your server rack,” he says.
To begin, the IT team needs to interview business departments in order to determine how important the availability and performance of particular applications are. They also need to determine what those departments are willing to pay to assure high uptime and performance if needed, he adds.
The idea behind self-service is that IT departments are often the bottleneck when it comes to deploying new applications or servers. Today, there are a variety of tools that provide businesses with the ability to better manage this process, significantly cutting down the demands on the IT staff as well as the time required to address end-user needs.
Self-service, Humphreys says, allows users to basically build their own VMs based on their unique — and often temporary — needs. For example, through the Citrix automated lab management tool, users can place a resource request via a web portal for server space, memory or processing capacity. In the physical world, he says, that process might take weeks or even months to fulfill.
But in a self-service virtual environment, that request can be handled within 30 minutes to an hour, assuming the user has pre-approval. Self-service provides a way to manage these requests, benefitting not only users but also IT staff struggling to keep up with technology demands.
Storage is sometimes the last component of IT to which businesses apply virtualization or consolidation. Perhaps that’s because data is often the fuel that powers the organization, and if data is lost, corrupted or unavailable, everything comes to a halt.
The good news is that storage architecture can be economically designed to support a company’s virtualized server environment. In fact, strategic application of today’s storage tools can lead to scalable and efficient virtual server storage.
For example, VMware's virtualization technology now offers dynamic storage to decrease the size of virtual machines. Significant savings can be gleaned from using the technology to create a dynamically expanding VM disk. Multiple VMs can be provisioned from a single image, adding even more cost savings.
Network storage may be an option for medium-size companies. If budget dollars are available, a Fibre Channel storage area network (SAN) is a way to centralize storage. In addition, greater efficiency and cost savings can result from thin provisioning, using storage only when necessary and investing in only a few hard drives.
As users become more comfortable with the complexities of managing a virtual environment with their servers, a next natural step is to move to the client device. Client virtualization separates the PC desktop environment from a physical machine using a client-server model of computing.
Client virtualization has been on the horizon for some time, says Richard Csaplar, a senior research analyst for the tech analyst firm Aberdeen Group. The model stores the virtual desktop in a remote central server, thereby offering multiple benefits, including greater security, fast provisioning, application availability, lower hardware and software costs, easier administration and centralized control.
There are multiple variables involved, Csaplar explains, and different models that benefit different types of users. One starting point, he notes, would be to adopt client virtualization for staff doing repetitive daily functions. The more basic and limited the applications involved, the easier it is to virtualize such client environments.