Introduction

Ever since humankind first made the decision to attempt to apply server virtualization principles to end-user desktops, there has been no end to articles proclaiming that “this year is the year of the virtual desktop.” Over the years, there have been various and sundry benefits and challenges identified around VDI, and there has always been optimism around the positive outcomes that can be had with the technology.

The year of the virtual desktop isn’t going to happen

At this point, we’re not going to see the “year of the virtual desktop” ever happen. If there was ever going to be a major and concentrated push to see VDI introduced into organizations, it would have happened at the same time that the "bring your own device" (BYOD) trend took place.

However, that doesn’t mean that virtual desktop projects aren’t happening. Instead, it just means that they’re happening differently than may have been envisioned when VDI hit the scene as a viable alternative to physical endpoints. Back then, pundits and vendors made over-the-moon claims about VDI, from promising all kinds of cost savings to operational nirvana, but soon, the cold harshness of reality set in and people discovered that VDI wasn’t going to be the silver bullet. Among the many issues:

• Discovery that server virtualization and desktop virtualization are very different beasts.
• Unexpected storage performance issues.
• Hidden licensing costs.
• A user experience that many found unacceptable.

Today, these issues have been largely solved, although licensing can still be a challenge.

The hype cycle

Although I don’t always like Gartner’s research, I do like the company’s Hype Cycle. Applying Garter’s hype cycle to VDI, I think we’ve truly hit the plateau of productivity or we’re at least on our way to that plateau.

As such, VDI isn’t the big splash that it used to be. There are a myriad of solutions on the market that can address desktop virtualization and there is a realization now that VDI isn’t always the desired endgame for the desktop environment. Instead, organizations have come to realize that, as is the case with server virtualization, making significant changes to the way that endpoints are deployed will depend on a variety of factors and can’t be attacked with just a single strategy.

A multipronged approach

The term “virtual desktop infrastructure” is giving way to “endpoint virtualization” as organizations undertake various approaches to changing the way that employees work. In addition, organizations are taking a more application-centric view of the environment rather than a device-centric one. With original VDI implementations, the focus was very much on the hardware and device side of the equation when, in fact, it really needs to be on the application.

And that’s leading to what we’re seeing in the market today. Today, there are many options from which CIOs can choose their endpoint virtualization path, including:

• Do nothing: For some users, this is a viable approach, particularly for those that have truly unique needs and platform requirements.
• Session virtualization aka Remote Desktop Services: Too often, RDS is dismissed as a “has been” technology when, in fact, it can be the perfect solution to a problem. In fact, I recently advised a client that was having major issues with a mission-critical application to use a combination of RemoteApp – RDS’ single application delivery service – combined with VMware’s ThinApp to solve the issue. Before making the recommendation, I tested the configuration myself to ensure that it would accommodate the application’s nuances, and it did so perfectly.
• ”Traditional” VDI: With VDI solutions that were on the market at the onset, the goal was to simply virtualize the desktop and place the resulting virtual machine in the data center and replace the user’s full desktop with a thin client. In many cases, this will still be the preferred method for handling endpoints, particularly those for general desktop users that may not have many unique requirements and that just use a handful of traditional applications. On the hardware front, many vendors have popped up providing specific guidance around using their solutions to handle VDI, so there is ample choice in today’s market when it comes to deployment platform – roll your own, go hyperconverged (i.e. Nutanix, SimpliVity, Scale), or use a reference architecture from a storage provider, such as Tegile or Nimble.
• Turn to the cloud: Several cloud providers over the years have begun selling desktop services, but with Amazon’s foray into the space, it’s become a truly validated way to approach the endpoint. In this scenario, Amazon provides the underlying operating system (Windows 7) and applications (depends on subscription, but can include Office), and users then use a device of their choice (PC, Mac, iPad, Kindle Fire, Android tablet) to connect to the desktop session. Amazon has partnered with Teradici to provide PCoIP to Amazon Workspaces users. PCoIP is a protocol that enhances the multimedia experience for the end user. These kinds of services make it possible to deploy virtualized endpoints without having to go through the process of building what can be a very complex data center environment.

Endpoint virtualization is just another service now

Given the variety of options available and the various different vendors that participate in the space, I don’t see any year ever being proclaimed as the “year of VDI” anymore. Instead, I see desktop or endpoint virtualization as just another arrow in the CIO’s quiver and for which the CIO needs to make ongoing business decisions based on application needs, not on architectural purity.

Action Item: Even though VDI may not get the massive attention that it once did, the endpoint virtualization concept has matured sufficiently so that CIOs can and should look at all of the options before them when deciding how to move forward with application delivery in a modern business. VDI alone, in fact, shouldn’t be the sole focus. Instead, consider all of the options and apply each one where it makes sense.

Footnotes: