VMware Horizon

Rightsizing Your Approach to Desktop Virtualization

Over the last few weeks I’ve travelled in North America and Europe, meeting with both VMware customers and many of my new colleagues. The only thing uniting all of these is a common realization that virtualization technologies have a key role to play in the End-User Computing (EUC) journey.

No two customers are the same: each has unique requirements and each is at a different point in their own journey. This is no surprise to me, given the many thousand of discussions I’ve previously had with organizations about their EUC strategies. What has taken me aback is the occasional publicity given to the apparent “failure” of desktop virtualization projects – this is far removed from my own views and the opinions I’ve heard expressed by our customers. My curiosity drove me to look a little deeper into such stories and I reached two simple conclusions:

  • the common theme to “disappointments” was over-ambitious deployment plans
  • organizations that set out to shift 100% to virtual desktops are the most likely to be disappointed

In other words, it was not that the projects were failing, but that initial expectations had been set too high. Let’s look again at what we know about the most popular (and mature) category of desktop virtualization – (Server) Hosted Virtual Desktops or VDI.

Hosted virtual desktop (HVD) products like VMware View move the traditional thick client user environment from the desktop to the data centre. The desktop software is managed in the same way as before, giving investment protection for all of the critical management processes and tools that organizations rely on to keep users productive. Since the desktop images are centralized and standardized, the way these processes and tools work can be made much more efficient. In other words, the cost of operations can be reduced – in some cases significantly.

It is this efficiency that makes VMware View so appealing: in financial-speak, HVDs offer diminishing marginal costs in operations (whereas as traditional PCs deliver constant marginal costs). By moving users to View, you can reduce operational costs and the more users you move, the more you can save. For those that deploy, that’s a strong motivation for including as many users as you can in the project – and there’s the rub, because HVDs don’t work equally well for all users and/or applications, yet.

With each new version of View, we have extended (and continue to extend) the range of applications and users that fit well. Although it was VMware that first supported the HVD architecture and offered an HVD product, we are in a market and the other HVD products have done (and will probably continue to do) likewise.

Improvements in how users access their desktops have significantly expanded the range of applications supported: our PCoIP protocol already works well with rich media applications and some of the exciting developments we have in the pipeline will take that to the next level, but separating the user interface from where the application runs necessarily imposes constraints, particularly if the user is a long way from the data centre. The way that View manages desktop images is at the core of the operational cost savings, but relies on standardization of desktop software. As we modularize the way desktop images are built and stored, we are extending those efficiencies to more diverse requirements, but we are not yet ready for all.

In other words, the key to success with HVDs is “rightsizing” the deployment objective for current technology capabilities. When used to support the right group of users it will deliver all the efficiencies and benefits of centralization as expected, but if you stretch it too far and too quickly you risk running into issues – and as ever with desktops, remediating those issues will bring new costs and complexities.

As desktop users, we are not all the same. No surprise therefore that we cannot all be addressed by the same technology solution. It is this diversity of requirements that makes each organization’s EUC journey so unique.

If it sounds as though I’m passing the buck here, that’s not the intention – actually, I’ve yet to see View at the heart of one of the “disappointing” deployments (having met VMware people in the field, I suspect this has something to do with the skills of our System Engineers). What I want to do, however, is remind you that your EUC future is probably much more diverse than the past. Everybody might have had a PC previously, but in years to come you will put EUC capabilities in to users’ hands in multiple ways. In other words, there is unlikely to be a new “one size fits all” solution that meets all user needs.

Let me close with a thought from one of our customers – as ever, the smartest people we meet: “with all this new choice in EUC technology, it’s getting really confusing and the decisions we have to make are getting more demand driven”. That can be quite daunting, but it’s also a good thing too as it gives you more control – provided you know what you want. But then it is your EUC journey: we are just making sure we give you the support you need on the way.