This content is part of the Essential Guide: VDI pilot project guide
Manage Learn to apply best practices and optimize your operations.

Virtual desktop architect shares tips from the trenches

A VDI architect offers words of wisdom he wished he knew about desktop virtualization, and offers IT pros seven tips to a successful desktop virtualization build-out.

Experienced VDI users understand why so many desktop virtualization projects never get past the proof-of-concept stage; it’s a complicated technology full of stumbling blocks with very little best-practice guidance from veteran users. So, as an IT architect of a global company that has used virtual desktops in a private cloud since 2001, I'd like to share some of the hard lessons I’ve learned, which you can use to build a virtual desktop infrastructure.

Get ready for some reality checks.

1. ROI is not the reason for VDI
If you have done your homework, you've already discovered that the return on investment (ROI) promises of virtual desktop infrastructure (VDI) aren’t being realized. You may be wondering, "Then why should I move forward with VDI?" There are some practical reasons to move away from traditional PCs to a virtual desktop environment.

One reason is that "bring your own PC" (BYOPC) is coming and you can't avoid it. Just ask the IT help desk administrator who had to get corporate email working on hundreds of employee iPads last year. He'll tell you that the next-generation workforce is here.

That said, the solution for an efficient, mobile workforce is not in the portability of the device but rather the portability of users' desktop experience. And since VDI allows access to applications from almost any device, it is the environment that enables BYOPC.

In addition, I predict that and will sell cloud desktops to consumers and small businesses by 2012. As more cloud-based services become available to the general consumer, IT will need to incorporate them into the computing environment. VDI is a natural way to do so.

2. Carriers don't understand VDI
History has taught us that applications motivate innovation on the network and not the other way around. We saw what Web browsing did for the proliferation of Internet connections and what Voice over IP did to raise the need for quality of service-enabled networks. The game Doom even had an effect on my own home network in 1994 (and at work; don't tell my old boss).

So when preparing your WAN for VDI or looking to deploy a new carrier, remember: legacy bad, innovative good.

As the virtualization community grows, demand will drive innovation in the networks. Keep your eye out for the new breed of carriers like Masergy Communications Inc., and Exponential-e that employ innovative designs with modern technologies.

Also, keep in mind that real-time computing environments need real-time monitoring. Carriers like Sprint and AT&T provide reports and analyses that are 15 minutes old or even older. A VDI environment needs monitoring that provides information no older than 60 seconds.

3. Monitor the user experience, not the system
Anything that affects your VDI environment will always affect the end-user experience and user satisfaction with virtual desktop, so you have to monitor everything that affects your virtual desktop environment. Here are some recommendations:

  • Employ a robust set of monitoring systems.
  • Use synthetic transaction monitoring from virtual PCs at remote branches.
  • If you run a Citrix VDI environment, deploy Citrix EdgeSight, which measures performance from end to end.
  • Take time to actually watch the busiest users during their most intense periods to understand the experience of your users.

4. The simplest application is not your simplest challenge
I'll keep this one simple because it’s a simple point: Don't ignore the simple applications used by the "simple" users. Just because an application is not resource-intensive does not mean it does not deserve your focus. On the contrary, simple applications are easy to use, memorized by users and usually result in fast keyboard transactions. High keyboard transaction rate, plus high latency, multiplied by VDI equals a highly frustrated user. Is that simple enough?

5. WAN optimization can help
Three years ago, I met with Riverbed Technology product managers, who told me that the company did not accelerate VDI (specifically Citrix ICA) because it was already efficient, compressed and optimized. I had a simple argument for them: Imagine 100 people in China receiving the same email from marketing to open the same intranet page in their virtual desktops in California. What if that image could be cached by the first user to a local Riverbed device? Following my advice, Riverbed accelerated ICA a year later.

Don't take WAN optimization off the table just because you are moving to VDI. You need it even more for a few reasons:

  • Document outputs (scanners, printers, etc.) will traverse the same network.
  • File transfers will still take place, especially for laptop users.
  • Large sites will share common screen images (as in the China example above).
  • All of these will fight against the user's virtual desktop experience.

6. You are alone
I hate to tell you this, but you are alone. If you start a VDI user group in your area and invite only those who have fully deployed VDI, make sure you buy just two doughnuts. The first one is for you, and the second one is to console you when no one shows up.

While at Simpson Strong-Tie last year, I searched virtually every continent and found few who had experience in an environment comprising mostly virtualized desktops. One was headquartered in Texas, and the other was in the U.K. We would have organized a user group meeting, but I don't know what a doughnut is called in England.

So, find a good online community to gain valuable information from peers, then hope that a local one starts soon.

7. You need to build an internal VDI team
Since just about everything in enterprise IT can affect a virtualized environment, VDI requires cooperation from separate IT teams. With that in mind, here are some suggestions for staffing:

  • Get the desktop support team trained on the concepts of VDI. Do you really want your VMware server administrators managing and supporting virtual desktops?
  • If people don't play well in the sandbox now, wait until your first major VDI-related problem to see a fistfight. Get your IT engineers to understand the need to design, deploy and support the VDI environment as a team.
  • VDI must be an initiative driven by -- or at least heavily supported by -- the CIO and other executives.
  • Let the help desk team determine user satisfaction and issue closure. They know the users best and therefore can access the VDI experience well.

A VDI project may seem daunting, but the community is growing, the benefits are evident, and IT will gain efficiency. Be realistic about the challenges to set expectations, then drive forward with your VDI plans. The few of us in the community are looking for others to share in the pains and successes.

Eugene Alfaro
leads IT Engineering for Cornerstone Technologies, an IT engineering services firm in San Jose, CA. He has architected, managed and operated corporate IT environments for multi-national companies since 1998. He has been a speaker on topics such as virtualization, WAN optimization, enterprise storage, Voice-over-IP and others.

Dig Deeper on Virtual desktop infrastructure and architecture

Join the conversation


Send me notifications when other members comment.

Please create a username to comment.


Tank you for your ROE (Return on Experience)interestiung article ..

You're not alone !

Here, at the Rennes 2 University in France we put a VDI project on the road in 2006...
It's was entered in production stage in 2008
Actually is up and running and it count 1,000 VM & Wyse thin clients. We use Vmware VSphere, Leostream connection broker, Citrix provisionning server for desktops, Wyse's thin clients and HP blade servers.

Your strongest argument seems to be the BYOPC undercurrent.

It may be true that this will come to realisation to a larger scale in the future but if I were a betting man i'd say that web application will reduce if not eliminate the need for large desktop for most user.

That leaves me with your ROI statement... BYOPC or VDI itself, in the end, it's always about ROI or business value. If not, it should be.

Little ROI, much complexity, low market penetration, the same business value can be achieve with other, cheaper, alternative... Unless I'm one of the rare niche case that would benefit, it won't be very high in my priority list for at least another 2 years i'd say... Then we'll see...