Will VDI decrease or increase your desktop TCO?

What's the top benefit you expect to get when you virtualize your desktops? I'll bet most of you will say you expect to substantially reduce the cost of ownership. And that makes a lot of sense, as desktop management is one of the largest cost centers in IT. The approach to managing desktops has changed little in over a decade, making it ripe for improvement.

According to Gartner, the total cost of ownership of desktops is dominantly operational costs (human time), and not capital costs (hardware and software). That's a good thing for VDI, where vendors struggle to get costs even close to physical desktops and laptops. So the question is, does VDI substantially improve operational costs?

Non-persistent desktops to the rescue

Why is desktop management so expensive? One big reason is that desktops degrade. They start fresh and new and wonderful, and over time get cluttered, confused and slow. So an obvious solution is to make the desktop brand-spanking new each time we use it. Which is exactly what the non-persistent VDI desktop does.

For use cases where everybody can use a homogenous desktop (one that is exactly the same for everyone, and can be flushed out after each session), the benefits are substantial. Desktop not working? Just restart it - you will get a brand new one. The savings potential is tremendous - some IT departments report a 20% or more reduction in help desk and desktop management costs. No level 2 or 3 escalation to fix a desktop - just get a new desktop.

But how many users can adopt non-persistent desktops? This remains a deeply debated topic, with extremes from 90+% to none. The truth is that it depends on too many factors for there to be a "golden rule". But it is well recognized that there are many application needs and personalization requirements, and trying to shoehorn them all into a single desktop has proven impossible. There are many reasons why - from users that need administrative rights (developers, power users, ...), to the high cost and complexity to trying to virtualize all the departmental and user required applications (see How many applications have you virtualized?).

For this model to be successful, you need to be able to have a substantial number of desktops using it. If only 20% of the users can use a non-persistent desktop - then 80% of your costs remain unaffected. That's why VDI is most successful today in the largest of organizations, where they have substantial call centers and task workers that can benefit from this approach - where saving opex costs on even a small fraction of the desktops turns into big savings.

Persistent non-persistent desktops

The latest trend in non-persistent desktops is adding a layer of persistence to them. Technologies such as RingCube (acquired by Citrix) are now providing an ability to add a persistence capability to non-persistent desktops. Now, when users install applications, change settings or do anything else to their desktop, the changes will persist ... they will be there the next time the log in.

The desktop itself is still non-persistent, but what does that mean now that all changes persist? The benefits of easy patching and provisioning desktops generally remains, but the big opex savings of getting a new desktop have disappeared - because the desktop isn't new and fresh each time you get one. It will age and get crufty just like before. And before long, even patching the base O/S and apps in the gold image will start to fail, as settings and executables in the personalization layer start to conflict ... resulting in falling back to using traditional desktop management tools and expensive level 2 and 3 support to debug these problems. And things get even worse when you install departmental applications in individual desktop persistence layers, as those end up being no more supportable than before virtualizing.

Complexity ... the enemy of savings

Why is this all so complicated? It's no surprise when you consider all the different, disconnected components involved in delivering the desktop ... operating systems from the broker, applications from app-virt, user and departmental apps in a personalization layer (and that's a simple view - often there are 5-6 different management consoles to operate VDI). There is no overarching and integrated management of these components, so no surprise that the opex benefits are elusive.

A desktop for everyone

Unidesk takes a different approach. We deliver every aspect of the desktop in individual, tightly integrated layers. One layer for the operating system, multiple layers for the corporate apps, one for more layers for departmental apps, and two layers for the personalization. Instead of different technologies at each of these levels, one holistic solution delivers the entire desktop. Operating system needs to be patched? Do it with a single image across all desktops. Need departmental apps? Put them in layer - manage once, deliver many. Users need administrative rights? Or special one-off apps? Install them directly on the desktop. No need to sequence, package and virtualize your apps just to support them - just install them into a layer once and you are good to go. There are immediate savings just in terms of simplicity and the ability to deliver desktops across the vast majority of your desktop use cases.

Unidesk holistic layering

But the real savings come when you look at the benefits of a integrated layer solution. By delivering two layers for user personalization, we are able to separate user installed apps and settings (that might break a desktop), from the valuable documents and data. We then automatically snapshot them on a regular basis, allow for complicated desktop problems to be repaired simply by rolling back the applications and settings layer to a recent but prior point in time. All the documents and data remain, and whatever damage the user may have done to the desktop is removed. You might think of this as surgical non-persistence - only remove the damage, not the data.

Maybe even more powerful is our ability to repair a broken application. Let's say Word is failing to work. The only way this can happen is something has changed in the user's persistence layer - as all other layers are non-persistent. You could roll it back, but even easier is to just tell Unidesk to "repair Word". We automatically examine the layer that has Word, and eradicate any conflicts that are found in the users persistence layer. Like magic, a level 1 help desk person can solve a problem that would have required hours and escalation to level 2 or 3.

Real world savings

Want proof? In Ohio, the State’s Department of Developmental Disabilities is 2/3rds through a 1,500 desktop conversion to VDI. With Unidesk, they’ve been able to reallocate 7 of their desktop support staff to finally address their more strategic IT project backlog. In California, Menlo College is in the process of virtualizing all desktops on campus. It was taking Menlo IT staff a combined 800 hours to patch & update their PCs each year. With Unidesk, it's 125. And in Wisconsin, the CIO of the Department of Children and Families recently said "We've realized a 30% savings for workstation support, and about a 40% decrease in call volume to our help desk" since implementing Unidesk and VDI.

One management console. Simple application packaging and delivery. Holistic layering. Innovative desktop repair. Support for nearly all use cases.

That's how you find real opex savings with VDI.