Whither VDI

My second article for Real User Monitoring, almost a year ago, closed with consideration of virtual desktop infrastructure (VDI), a “thin-client” approach to management of what end-users see. It’s time to return to VDI and its latest trends.

The main trend is that there is no overwhelming trend. At least a few serious observers once predicted that VDI would take over information technology (IT) operations. It has since become apparent that VDI’s destiny is more middling than that–it’s in use, and continues to expand, but there’s no prospect it will exterminate conventional PCs. Now, VDI looks like just one of many technologies IT architects need to juggle for the best results. ┬áVisibility into the end-user experience, however, continues to be important in other ways, as seen through the use of today’s monitoring tools.

To understand VDI as a technology it’s essential to see its business model clearly. VDI has been associated with several strong, aggressive, but medium-size vendors, including VMware and Citrix. As this series often mentions, there’s a skew between what vendors can effectively monetize, and engineering optima. VDI often bundles Microsoft operating-system licenses and/or proprietary networking protocols which combine technical and business value in variable measure. Analysts regularly judge VDI in what Gartner calls the “gestation” phase. Although several documented cases report a VDI deployment as a big win, such projects appear still to depend on specific trade-offs of licensing, convenience, and manageability.

One improvement in the VDI marketplace is that ‘desktop-as-a-service‘ (DaaS–often called ‘PC-as-a-service’ a few years ago) has matured. It’s now easier and less risky for small organizations to purchase turn-key VDI on a pay-as-they-go basis, and leave server-side headaches to experts.

There certainly is plenty to know on the server side of VDI. A Gartner analyst estimates storage requirements, and networking is also a substantial commitment in any VDI project.

Still, most of what’s visible about VDI resides in the client device. That’s where end-users perceive quality-of-experience (QoE), and where most VDI-specific innovation takes place. One of the important QoE lessons is that traditional measures of system “burden”, including networking load and storage latency, do not entirely determine what an end-user observes when moving the end-point mouse. Any successful project must adopt at least a few techniques from real user monitoring and performance management to ensure that end-to-end responsiveness stays in bounds.

Zero-client experiments

Another variation on VDI still growing in importance is so-called “zero clients“, that is, end-user-level hardware whose only responsibility is VDI. These clients can be so skinny as to only need a single power-over-Ethernet (PoE) connection, in place of separate network and power cabling, which makes for a tidier and less-expensive workplace. One customer, with Miami-Dade County municipal government IT, quotes, “This device uses seven watts, versus 150 to 300 watts …” As enthusiastic as I am about low-power computing, this overstates the gains, because comparable modern fat-client equipment should be under 100 watts, while most zero-client hardware falls in the range of 5-30 watts. Chromeboxes, with which I continue to experiment, typically tick along at about 10 watts.

Strategically, VDI has often been presented in terms of cost savings and liberation of IT from mundane chores of desktop maintenance. These gains shrink into the background, though, for organizations that are able to co-ordinate VDI with mobile strategy. “[T]he ability to effectively deliver business applications across stationary and mobile devices“, as Arthur Cole explains it, will eventually trump everything else described here in the near future of mobile’s dominance.