Step 3: Analyze Total Cost of Ownership
While desktop virtualization can result in cost savings when implemented judiciously, these cost savings are not guaranteed. Any organization considering desktop virtualization should closely analyze the true costs of implementing and maintaining a virtualized desktop environment. To be useful, any total cost of ownership (TCO) calculation needs to account for all material impacts to the TCO; additionally, the inputs and assumptions of the calculations must reflect the realities of the organization. While vendor-provided TCO and return on investment models can be useful, use them only as a starting point for a more detailed and comprehensive TCO analysis that takes into account the following seven factors: Hardware Software Energy Data center real estate Support Implementation Usability The remainder of this section provides details on each of these factors. Hardware Reducing hardware costs is often the driver for organizations to implement virtualized desktops. By moving the computing power to a central and shared location, users who formerly used intelligent desktops can be provided a much lower-cost thin client device to access the shared computing power. However, estimates about the number of virtual users who can share a single server vary significantly—from 6 to more than 70. According to a 2010 published survey from Gartner, most organizations average 6 to 8 users per server, and very few organizations manage to deploy more than 20 users per server.1 Four primary elements affect hardware costs and the number of virtualized desktop users that can be successfully deployed per server. User type: User type has a significant impact on the amount of hardware required to successfully virtualize the desktop environment. Users who infrequently access the system or access only small, lightweight applications can share a server in higher numbers than “heavy” users who spend the majority of their day accessing graphics-intensive or analytical systems. Consider deploying different strategies for each user group: for example, virtualizing “lightweight” users at a different ratio of users per server than knowledge users, and keeping power users in a traditional intelligent desktop environment. Usage spikes: Another key consideration, often overlooked, is ensuring that the environment can handle spikes in usage as well as day-to-day operations. Usage spikes often occur at key points in sales and financial reporting periods, during key seasonal events, or during “one-off” events such as new product launches. If hardware capacity is planned without considering usage spikes, business may run seamlessly 80 percent of the time but experience significant slowdowns during the business’s most critical periods. Server size: The size and type of the server has a big impact on the number of virtual users that share a physical server. The important consideration is not the number of servers but rather the total cost to procure, manage, run, and store the servers. Larger, more powerful servers will have a higher up-front cost and cost more to cool, but will hold more users, take up less server floor space, and cost less per virtual user to support.
1 Dayley, Alan. User Survey Analysis: Virtualization Cost Reduction and Consolidation Promises Slow in Coming. Gartner, 2009
13 Intel IT Center Planning Guide | Desktop Virtualization