Virtual Desktop Infrastructure (VDI) is a term coined by VMware for a specific type of desktop virtualization. Users run their desktops on a virtual machine that lives on a centralized server within a datacenter. It is an extremely efficient type of desktop virtualization as it gives users a cost-effective service, which contains both the simplicity and optimal security of centralized management. This centralization gives users the ability to access their desktops when, where, and however they want; across the globe, and on any device. (If this is a little unclear, see our introductory article on this topic.)
VDI has received a bad reputation over the past couple years due to a perceived lack of innovation. One of the biggest complaints about VDI is the way that a mouse can sometimes lag as an end-user drags it across a screen. Many times, a user will try to enjoy a rich media presentation and the virtual machine seems to bottleneck, making it seem as if the virtual desktop is underperforming.
These types of performance bottlenecks aren't always due to internet connection. While it's true that ping time and geo-location as well as your local ISP have some say in the way services are delivered, a well-thought-out plan for an organization should already have most of these concerns addressed before building out this type of infrastructure.
Examples of some of the more impressive applications for this include CAD and 3D art. Both need a graphical processing unit, more commonly known as a GPU, in order to function correctly. In a traditional cloud VDI deployment, the core CPU tries to emulate a GPU. The CPU isn't meant to do GPU calculations and this is why you can see lags in rich media presentations. The solution is to implement a cloud GPU card in your private datacenter or to call your public datacenter and ask them if virtual GPUs are offered in their lineup of services.
Setting up virtual GPUs in your private cloud could be costly because the technology is rather new. If the types of applications your users utilize require virtual GPUs, the money you are spending is probably being well spent. Servers such as the HP DL380z now feature the Nvidia GPU GRID technology in its 2U design. Pricing for this server has not yet been announced; however, this server will allow for high-performance desktop computing inside of a private cloud—which is exactly what many organizations are striving to accomplish.
Could GPU virtualization be the catalyst for changing the perception of virtual desktop infrastructure? Once end-users become accustomed to the reliability of using a virtual desktop inside of a high performance cloud, the perception of VDI services should increase. Being able to thoroughly plan out this deployment is the key to success. For example, if you are building out a private cloud using servers such as the HP DL380z, you should theoretically keep those servers as close to the end-users if at all possible. This will allow you to craft internal service level agreements that ensure end user satisfaction. Increased end-user satisfaction goes a long ways towards changing the perception of virtual desktop infrastructures.
About the author:
Natalie Lehrer is a senior contributor for CloudWedge. In her spare time, Natalie enjoys exploring all things cloud and is a music enthusiast. Follow Natalie's daily posts on Twitter: @Cloudwedge, or on Facebook.