VDI can be cheaper than physical desktops, but you'll have to give up some desktop functionality if you want the virtual instance to be evenly priced.
Last week I wrote about how cost models aren't sufficient for determining the overall cost of a VDI project. Virtual desktop infrastructure (VDI) is a complicated technology with many parts, and any cost model will surely have to make generalizations that glaze over the things that are important to you (like power and cooling requirements) or place too much importance on certain things (like power and cooling requirements).
Whether that's done accidentally or on purpose, the general goal is toward low-cost VDI -- showing that desktops from the data center are going to be cheaper than physical desktops under cubicles.
Despite cost models' predisposition toward inaccuracy, it is possible to deliver a virtual desktop for the same amount of money as a physical desktop -- less, even. The problem is that comparing VDI to physical, and saying physical is cheaper, is like comparing a Ferrari to a Ford Focus and saying the Ford Focus is cheaper. Of course it is, but it's not exactly the same thing, is it?
It comes down to capabilities
The typical $600 physical desktop today has a processor in it with four cores, a 3-D graphics card, the ability to drive two 1920x1080 high-definition displays, HD webcams, 3-D sound and any number of crazy USB devices. You may be shaking your head and saying that your users don’t have that kind of stuff at work, but think about what they are using at home, too. There they could have 8 cores, four displays, and USB devices so crazy that the ones they want to use at work make you shake your head again. (Seriously, there’s a lot of them).
VDI, though, is focused on density, and the way you measure density is through a metric called "vCPU's per core." That name alone should indicate the problem: You're trying to see how many users you can fit on a single core of a processor. Physical desktops are about throwing as much power into a small device as you can to hit a certain price point -- regardless of whether or not the user actually uses that power. Virtual desktops, on the other hand, are about fitting as many desktops onto a piece of hardware as possible, leaving little to no excess resources.
Gone are the four cores the user had at their disposal with a physical desktop. In order for VDI to work and save money, that user now gets one-eighth of a core. The same reductions happen in sound and video performance. Two screens at HD resolutions take a lot more back-end resources than one screen at good-enough resolution, so now it's just one screen at 1366x768 for your unfortunate user. Sound, USB device support and webcams all come at too high a cost, so every aspect of a user's desktop takes a hit when moving it to the data center.
More on VDI cost
How to calculate ROI from VDI
Benefits and challenges of VDI
Finding those hidden VDI cost savings
The end result is a virtual desktop for around the same price as a physical desktop. Of course, with this kind of low-cost VDI, this desktop has just a fraction of the functionality the user has grown accustomed to. Add to that the lockdowns and restrictions put in place to make the desktop more manageable, and the end user experience changes drastically.
That's where the biggest argument surrounding low-cost VDI comes in. If you were to compare apples to apples, functionality to functionality, there is no way that we can provide virtual desktops for the same price as physical desktops. To provide physical desktop capabilities for each VDI user, it's going to cost money. You have to buy hardware, a storage solution, networking, and you have to invest in the endpoints.
So, is it worth going that far? That depends on your goals.
This was first published in July 2013