May 2008

Increasingly, IT organizations are leveraging virtualization capabilities to improve hardware utilization and lower operational costs in the datacenter. Virtualization is being used to provide dynamic and flexible allocation of infrastructure resources to applications and business services to meet service-level requirements. Strong growth in this market will continue as organizations increasingly deploy virtual machine software (VMS) as a means of decoupling the application stack from the underlying hardware. IDC views VMS as a foundational technology for the creation of agile IT environments and expects robust growth in this market through 2011.

The following questions were posed by AMD to John Humphreys, program vice president with IDC's Enterprise Platform Group, on behalf of AMD's user customers.

Q. Where is virtualization in the market today, and what are the key drivers?

A.  IDC considers virtualization to be a mainstream technology in the market, and we find users to be pretty bullish about this technology. We attribute this to the fact that the return on investment in the business case is very compelling to users. As budgets are being pushed even harder by CFOs and CEOs amid the onset of a U.S. recession, virtualization provides a good way to reduce costs. For a software investment of $5,000, companies are able to eliminate about $25,000 in hardware costs. We also find that, on average, a quarter of the companies that have implemented this technology have already virtualized about 25% of their existing environment. These same companies expect that 50% of their environment will be virtualized within the next 12 to 18 months.

The reduction in capital costs associated with the acquisition of new server hardware is a primary market driver, along with the ability to reduce some of the operational requirements for power, space, and cooling. The virtualization market is being embraced by an initial wave of adopters focused on cost benefits. This includes encapsulation, where companies can consolidate multiple applications onto a physical server and avoid the inevitable complications of application contention and regression testing.

Q.  How are buyer motivations for virtualization changing?

A.  In the past, the motivation entailed leveraging the encapsulation attribute of the technology. After experimenting with virtualization software, customers like the fact that it's also mobile in terms of decoupling the application stack from the underlying hardware. As a result, virtualization software actually expresses the virtual machine — the "server"— as a file. When a server is expressed as a file, it has all the attributes of a typical file and can be copied, moved, replicated, and even emailed.

As such, virtualization software can be used to solve some business continuity problems, including a new way to do disaster recovery. Its flexibility allows IT to more easily manage a disaster recovery site. It eliminates the necessity of having mirror images of existing datacenters by extending disaster recovery benefits to a host of applications, regardless of whether the type of hardware is different at the primary and secondary sites.

We think of this scenario as bringing disaster recovery to the masses. No longer is disaster recovery reserved for protecting only the most mission-critical assets. Now, even companies with modest IT budgets can protect their assets and ensure data recovery in case of power outages, natural disasters, or other types of failure.

Its file mobility attributes and capability for live migration also allow users to take a live, running virtual machine and move it from one physical host to another. This also enables easier hardware maintenance and capacity planning at the resource pool or cluster rather than at the individual application level.

Q. What is Virtualization 2.0, and what are its benefits?

A. Virtualization 2.0 refers to anything that goes beyond strict server virtualization to drive consolidation, including use cases such as disaster recovery, business continuity, high availability, and similar uses for the mobility components of the technology. Companies are also beginning to work with the concept of virtual clients — so-called VDI deployments — whereby actual desktops that aren't mobile can be consolidated into the datacenter, run inside a virtual machine, and then accessed via a subremote graphics protocol such as RDP or ICA.

People are using this primarily for data security — as a way to avert the problems (and bad press) associated with the exposure or theft of customer information when drives or laptops are lost or stolen. They're also using the flexibility of virtualization as a way to increase user productivity. It's much easier and often quicker to provision a virtual machine with a desktop than it is to provision a new desktop to a user. For example, one customer told us it took about six hours to provision his users with a new desktop when their hardware failed. In the virtual world, it took about 15 minutes, which significantly improved the productivity of the end users because of the flexibility gains achieved through virtualization.

Finally, we believe virtualization will become a foundational technology to enable concepts such as "cloud computing." In our opinion, cloud computing leverages technologies such as service-oriented architecture in the sense that a component-sized application (e.g., service-oriented application or traditional 3D architecture) can be put in virtual machines and scaled on demand. Virtualization is a way to manage a business application from end to end, and scale it when required. It enables the concept of putting services in the clouds — somewhere outside the four walls of an organization. This won't happen anytime soon, and some significant challenges must be addressed before it becomes a reality. However, virtualization not only makes cloud computing possible but also provides a road map for getting there.

Q. What are the challenges to wider adoption of virtualization?

A. Many companies have told us it takes them somewhere between three and six weeks to get a new server up and running, from the time they purchase the equipment until it is shipped, installed, tested, and ready for use. In the virtual world, all this can be accomplished in a matter of hours — in some cases, even minutes. As such, it increases an organization's appetite for servers. For example, one organization was able to consolidate its systems down to 200 physical devices hosting 1,000 virtual machines. After about six months the company discovered it was actually supporting 1,300 virtual machines — an increase from the original 1,000. Since it was so easy to provision a user with a new virtual machine, IT was creating virtual machines as a way to resolve the demand from different groups, departments, and divisions within the organization to host new applications.

Although virtualization eliminates the cost of purchasing new hardware, it is much harder to keep track of virtual machines than physical hardware systems. Consequently, the number of virtual machines grew by 30% in six months. To address this situation, the organization started sending emails simply asking its employees whether the virtual machines they had been provisioned were still being used. The company then was able to decommission the idle virtual machines, reduce the number to about 700, and reclaim that capacity, making it available for other uses. Planning for decommissioning and archiving virtual machines over time clearly is a new process for organizations.

Additional challenges facing the industry and customers involve the coordination and management of virtual machines. First, companies have to learn how to track and assess the value and usage of virtual machines. In addition, since the virtual machine is mobile, security needs to live with the virtual machine. Data and network access also needs to be mobile, which means provisioning the network to the virtual machine and not the physical device. This entails a new set of corporate policies and IT collaboration to effectively manage the logistics.

For example, data networks need to know where the virtual machine resides; companies have to know how to move their security policies as the virtual machine moves; and companies have to ensure that the virtual machine has access to what it needs, including the appropriate data and storage network. As virtual machines proliferate, companies have to make technologies, such as the network, become more dynamic. Finally, the ease of migration of virtual machines from physical host to physical host depends on the environment and the demand characteristics for the application. This migration generates a significant amount of network activity that essentially has to be done manually. Consequently, there is no easy way to dynamically provision, rearchitect, or reconfigure the network.

Q. Is there a Virtualization 3.0 on the horizon?

A. Virtualization 3.0 envelops the concepts of service-oriented architecture and cloud computing. Although it is really just a continuum from 2.0, we see many IT departments interested in the concept of cloud computing as a way to more effectively manage the services that they provide to their organizations or to their customers.

Concurrently, executives within organizations view cloud computing as a way to change the IT model from a fixed-cost model to a variable-cost model, whereby they can pay for and charge back on a daily, weekly, monthly, or annual basis any service that's used or consumed by an individual within their organizations. This model is viewed by both IT and executives as a solution to address major concerns they have about servicing their organizations while trying to control finances, maintain high levels of service, or increase the levels of service.