Virtual Observer

Desktop Virtualization: What are You Waiting for?

In this new column, Taneja Group lead virtualization analyst Dave Bartoletti will offer his personal insights and findings on all trends virtual and cloud.

• By Dave Bartoletti
• 03/07/2011

(Editor's Note: It is my pleasure to welcome Dave Bartoletti, lead virtualization analyst at the Taneja Group, to the Virtualization Review fold, where he will be writing regular online columns on a variety of virtualization and cloud computing topics. I consider us lucky to have him, and I'm sure you will find yourself looking forward to his insightful "Virtual Observer" commentaries. -- Bruce Hoard.)

To start, I'd like to thank Virtualization Review for offering me a space to share the virtualization trends, technologies, and vendors I'll be covering this year. I've been steeped in virtualization for the last ten years and in my role as lead virtualization analyst at The Taneja Group I continue to be surprised and impressed daily by the pace of innovation in this exciting tech sector.

Take desktop virtualization. As 2011 kicks into gear, many in the vendor and investment communities are wondering (again) if this will be the year DV takes off. Of course, for many vendors, DV in one form or another has been flying high for some time, so why are we still asking the question?

There's a perception that DV has stalled, for many reasons: high storage costs for server-hosted DV (VDI), poor protocol performance (and user experience), limited successful use cases, security concerns and the vast array of client devices, peripherals, form factors, and application types, to name a few (well, a lot, actually). User workspaces aren't like servers, after all; they're messier, they often move around, and they are tethered to the datacenter via all kinds of slow networks.

The drivers for DV haven't changed, though; in fact, they're probably more pressing than ever. The cost of supporting all those new netbooks, smartphones and tablets is starting to overwhelm the IT managers we speak to, and many don't expect to finish their Win 7 migrations before Win 8 ships--hardware and OS refreshes alone are enough to prompt many to start DV projects. It is certainly cheaper (and safer) to patch and maintain a small number of golden desktop images, and virtualized endpoints are--at least theoretically--easier to secure.

When I talk to clients about their desktop virtualization hopes and dreams, several key themes recur. In order to ramp up DV, the vendor community must recognize and overcome them:

Too many choices. VMware has a mature VDI solution plus application virtualization, and targets its solutions at its loyal datacenter buyers. Citrix has an a la carte menu of options covering the entire sector, from VDI to application virtualization to a client hypervisor, and has lately begun to retool it marketing to appeal to corporate users as well, which is smart. Microsoft is all over the map, of course. IBM's going the Linux route, wisely, and is all-in on the cloud.

But none of these virtualization leaders has hit on the breakout approach that redefines how enterprises think about their user desktops, nor how users think about how they want their corporate data delivered. It's not an incumbent's game at this point--there's no unifying approach (like x86 server virtualization) to compare the virtualization leaders against one another, leaving the market wide open to start-ups with good ideas.

Shifting perspectives. Like it or not, the consumer device market will drive the next decade of end-user computing. Enterprise IT teams have traditionally created purpose-built apps and loaded desktops with a heavy OS full of productivity tools and locked-down data interfaces. But just like the corporate car and phone, corporate laptops and desktops will soon go out of fashion. It simply costs too much to deliver them.

The client device will drive the next wave of corporate data and app delivery decisions, and with each new device comes a new OS and a host of new compatibility issues. "Virtualizing" each app and OS into a new portable container to secure corporate data will drain even more resources, and isn't a sustainable strategy.

Clouds on the horizon. The Salesforce juggernaut has proven the SaaS model, and one of the most common challenges I hear around desktop virtualization boils down to: "Why virtualize the desktop if my apps are in the cloud?" Good question.

For many apps and many user types the desktop should fade away: corporate data held securely in the cloud should be delivered RESTfully to stateless endpoints, running thin interfaces on a user's own hardware. Almost all of us get most of our e-mail this way already. If you need PowerPoint, rent it and your company will reimburse you.

It's the management, stupid. The traveling knowledge worker use case we've just described isn't applicable everywhere, of course. There are still millions of desktops and laptops within corporate firewalls that have to stay there, in some form. Hospitals, brokerages, factories...all are bursting with in-house Windows desktops going out of date the minute they're unpacked. And for IT to get a grip on the costs of managing this infrastructure, it's going to come down to who packages up the broadest range of DV technologies and makes them all work from the simplest operations interface.

This is where we'll see the homegrown integration of VDI, connection brokers, app virtualization, profile redirection, thin clients and zero clients--all the elements of desktop virtualization--and this is where IT departments are aching for packaged solutions.

In my view, desktop virtualization isn't really about virtualization at all. It's about leveraging the consumer computing model we've created in the last 15 years to retool corporate computing. In my next posts, I'll dive into some of the exciting vendors and technologies--both well-known and emerging, established and start-up--that are poised to shake things up on the desktop. I'd love to hear from you: When it comes to desktop virtualization, what are you waiting for?