This content is part of the Essential Guide: VDI pilot project guide
Problem solve Get help with specific problems with your technologies, process and projects.

How to revive a failed VDI proof of concept

If you have tested VDI and the POC failed, you aren't alone. This article will help you assess the situation and provide guidelines to pick up the pieces.

You've heard about all the benefits of moving from PCs to a virtual desktop infrastructure and decided to give it a go. You negotiated to get the VDI licenses for free and finagled to have it installed at no cost. IT has even added "private cloud" to their vernacular. Unfortunately, your VDI POC went as far as you can throw the server it runs on.

What happened and how can you recover?

If this is your story, join the club. Many VDI projects fail for one reason or another. If you are about to start a virtual desktop infrastructure (VDI) proof of concept (POC), this article will help you avoid the pitfalls that many IT pros before you stumbled on. For those who have tried VDI and failed, use these tips as guidelines to pick up the pieces and turn it into the successful project you were hoping for.

Re-evaluate VDI assumptions
The first thing you need to do is ask why you embarked on the VDI journey. Were you proposing a quick return on investment (ROI) and/or a lowered TCO to reduce IT costs? Was it to shorten desktop delivery cycles? Was it to extend the life of existing hardware? Whatever the reason(s), IT justifications last about as long as the technology's ability to deliver on expectations.

Many articles written over the past year steer companies away from using ROI and TCO as justifications for VDI. Instead, look to some real-world applications that support VDI adoption. Here are a few:

1. VDI is successful in the finance and education industries because of the nature of their user base. The distribution of their user base makes support, system handling and maintenance difficult. In such environments, VDI can reportedly reduce support costs by over 50%.

2. Moving to Windows 7. Why deal with image distribution, multicasting and the headaches of hardware deployment? Using VDI, you'll have total control of the desktop deployment from anywhere to any device. The greater benefit is the ability to deploy changes in mass without physically revisiting the systems -- and the number of changes to a new desktop rollout is usually high in the early stages as issues are uncovered.

3. Bring your own device (BYOD or BYOPC) policies are becoming more and more common in companies today. VDI is a good way to deliver corporate apps to those devices.

Who to invite to the POC party
Have you ever attended a party that was more boring than watching penguins migrate simply because it was missing a few "life-of-the-party" friends? The same principle applies to POCs. Consider the users that you invited to test the system and assess whether they evangelized the concept.

Yes -- evangelism is required. A successful VDI POC is not one where you merely find the bugs and issues, but also where you excite the users about the new system they'll be using. Here are some types of users that are ideal for a VDI POC.

  • The "I-want-to-work-from-anywhere" user: These users are demanding and put in the extra hours. They are usually outspoken and have the attention of senior management. Use their strong voice to your advantage.
  • The "I-want-my-own-device" user: If BYOD is not a company strategy, don't try to get management support for it by yourself. Users can influence decision-makers with their collective voices and they can provide practical reasons.
  • The "I-want-my-app-access-now" user: One of the benefits of VDI and virtual applications is that it lets IT quickly deliver applications to users and empowers users with self-service provisioning.

You get what you pay for
If you wanted a free POC and it isn't going well, you got what you paid for. If VDI is a strategy supported by senior management, get some budget to hire a consultant with real-world VDI experience to help you with the design, concept and communication.

Most VDI implementations are done by VDI systems engineers who can setup and deploy. If you are still working to get business support for VDI, you need a consultant who has helped organizations successfully execute on the vision.

Get the team involved
Many VDI projects start with infrastructure and systems engineering teams because that is where virtualization expertise resides. But you also should include the desktop support and network teams. Though they may not have the virtualization experience, they have the most potential for understanding the value of VDI for the users (unless they think their jobs are on the line).

The user support group should be involved. If the user community likes their support group, then they are an untapped resource for spreading the word. If they are not liked, VDI may actually boost satisfaction scores by stabilizing the desktops through techniques such as non-persistent images the ability to support remote users. User support will also be able to manage desktops easier.

We normally don't think of the network engineering team as a beneficiary of VDI, but they are. One way it benefits them is by reducing LAN and WAN bandwidth to edge and remote sites. That's because the majority of network traffic happens between the client applications and the back-end servers. For example, when you use Outlook, there is a fat connection between Outlook and your email server. VDI allows you to put the applications closer to the back end therefore reducing WAN bandwidth. It also lessens CIFS traffic choking the network, improves remote user security and VDI simplifies VPN security policies.

If your POC succeeds, you'll need to get other IT teams involved in the project anyway. Hence, involving them early on gives them an opportunity to familiarize with the environment, which will reduce escalations to virtualization engineers when you get this show on the road.

Keep in mind that a VDI project is not a typical IT effort. It affects users directly and requires the involvement of more IT team members. While the benefits are well documented, you won't realize them unless you get full support and can execute it at the scale you are planning.

Eugene Alfaro
leads IT Engineering for Cornerstone Technologies, an IT engineering services firm in San Jose, Calif. 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. You can follow him on Twitter @Eugenealfaro.

Dig Deeper on Virtual desktop infrastructure and architecture

Join the conversation


Send me notifications when other members comment.

Please create a username to comment.

Great thoughts. I might add that it is wise to engage the Storage team, as VDI lives on storage (though it travels over the network.) Last but IMO, MOST IMPORTANT engage the Active Directory folks. Even (maybe especially) if you intend to use a Profile Management solution, you NEED TO KNOW and UNDERSTAND your GPOs to do VDI right.
The best way to revive a failed VDI proof of concept is to flip it to an IDV project. IDV, or Intelligent Desktop Virtualization, gives companies all the benefits of VDI including centralized management of virtual PC images. However, it executes the desktops on a secure local hypervisor on the desktop or laptop PC. This dramatically cuts the cost of infrastructure - storage, servers, and networking. More importantly, it delivers better performance as well as great mobility for laptop users.
Sham Sao, CMO
Most VDI deployments fail either because of the high datacenter costs or poor user experience.

We are working with a 15,000 PC opportunity that was evaluating VDI. The capital costs of deploying VDI for 15,000 users shocked them. Since VDI runs about 30 users per server, it would take 500 servers for 15,000 users plus quite a bit of high speed storage. The CAPEX for the servers and storage was approximately $3.5M. While that surprised them, the OPEX for those servers was nearly as shocking. It is approximately $800 per server per year for power and cooling of a server. Without even counting the power and cooling for all the storage and network gear, this equates to $400,000 per year in storage and cooling costs.

Users want to use multimedia apps, need to operate over a slow connection or disconnected from the network completely, and want to install their own apps. This is especially the case with PCs becoming more powerful for essentially the same price. This is certainly not the case with VDI.

What is needed is for IT to have all the manageability benefits of centralizing images in the data center and single image management of those centralized images, but at the same time enabling users to have the local PC experience that they need to be productive – all without breaking the budget.

Barry Phillips, CMO at Wanova
Great post, agree completely. VDI is as much culture/political as it is technical, and this has to be understood at the outset to avoid what you describe.

In our experience, following a structured methodology through POC, Pilot and Production deployments, not only mitigates the risk of experiencing this stall in the first place, but of course increases the chance of success, and increases the number of good VDI candidates identified. We live by our ‘Assess > Design > Migrate > Validate’ methodology, but when we come across these stalled POCs, they invariably have not conducted steps 1 to 3 correctly or at all. In that case we suggest starting with the Validate step to baseline where that POC is today, understand the user experience, and identify any bottlenecks. This takes the subjectivity out of the POC, and gives objective and credible data to support moving forward and/or fixing the broken POC. Then we start over again at step 1 to assess, design, migrate and validate the next set of candidates.

Ricky El-Qasem, Technical Director @ Liquidware Labs.
Agree with most..... We got it all right, nailed the UEM, mitigated the home-grown/custom apps, carved out storage, groomed the routes..... everything covered.

What we missed.... We didn't see the EMC train pulling in with a HUGE project while we were chomping at the bit with HDX.

Now I've got to re-assess whether to go with HDX or PCoIP.... What a pain!