Dickens enthusiasts might remember the scene in A Christmas Carol where the Ghost of Christmas Present reveals that ignorance and want are the foundation of society's ills. Similarly, VDI failure comes down to two true causes: misinformation and performance problems.
Yes, VDI projects can be derailed by diverse sources such as hidden costs, underestimated requirements, annoyed users or simply implementing the technology for the wrong reasons. But all of these causes fall under one of two umbrellas. Misinformation and performance issues can cause VDI problems to destroy your project before it ever gets going.
Many organizations simply lack the necessary knowledge to take the right approach to VDI; predictions and assumptions about VDI can be dangerous.
For example, some admins assume desktop virtualization is the same as server virtualization. Servers are generally predictable because they do the same things at the same times every day. Desktops are not predictable because users are not. Every user is different so it's impossible to know when he will watch a video, access an app that uses 3D graphics, require remote peripheral support or do anything else that causes traffic to fluctuate.
The differences between server and desktop virtualization don’t end with the user, however. The goal of server virtualization is to consolidate, but VDI does not consolidate desktop workloads; it moves them to a central location. You still have to manage the desktops, just like you did in a traditional environment. The only change is the location of the desktops. Also the disk I/O involved in VDI is a challenge unique to virtual desktops.
Organizations also tend to underestimate the hardware and network requirements involved with VDI. Hardware vendors offer information about how many users their product can support per core, which can give you a general idea of what you need based on the number of users you have. But until you use the hardware in your environment, you won't know if it can really support as many desktops as the vendor says it will, given your requirements and configuration.
From a network standpoint, VDI creates a new set of traffic to deal with. With virtual desktops, everything every user does goes through the network. If you don’t have enough bandwidth for the number of users you support, there will be delays.
Bandwidth problems usually have their foundation in faulty testing. If you test your VDI deployment on a small number of users and try to use those results to determine what your larger deployment will be able to handle, the results won't add up. Every user you tested might each use 1% of the space available, but others might end up using more or less than that once your implementation is rolled out. You could always throw money at the problem and try to improve network performance, but that can be very expensive, and will likely blow up your budget and take your VDI project with it.
Many organizations start out VDI projects thinking that non-persistent desktops are the best choice because they're less expensive and complex than persistent desktops. But the reality is that non-persistent desktops are too inflexible for some use cases, and users like to have their own personal desktops. Unhappy users can kill your deployment, but if you end up switching to persistent desktops, your storage costs will increase.
VDI costs often don’t present themselves until late in the game, so your cost models can unwittingly leave out important charges. Shops that implement VDI to save money quickly learn that the technology isn't usually a cost-saver. To prevent these pitfalls you must know why you're implementing VDI so you can set achievable goals to mark your progress.
Performance is key to a successful VDI project. If virtual desktops can match or exceed traditional desktop performance users will be happy. If performance dips, prepare for mutiny. Once you lose your users, it's hard to maintain your VDI project.
Remember, if a user thinks there's a problem, even if it doesn't appear to be a technical issue, it's a problem. To understand what's actually going on, you need a user's-eye view of the virtual desktop. End-user and performance monitoring tools give you that insight, along with availability and response-time statistics and alerts.
Hardware failures can compromise performance too. Hardware can give out at any time without the slightest warning. To safeguard yourself against random failures you need redundancy. If you have backups for each aspect of your VDI deployment you'll be covered in the event anything fails. Predictive analytics can also help by alerting you of any imminent failures.
Look at virtual desktop protocols, the ports VDI uses, subnets, VLANs or a combination of the four to ensure virtual desktop traffic is your top Quality of Service (QoS) priority. If QoS settings are mismatched or a device is not configured correctly, a router or switch could disrupt performance.
Storage performance issues caused by disk I/O latency can also get in the way of your VDI implementation. If you have the wrong storage type or location, users can experience more desktop or application lag. The best way to fight the problem is with faster hardware, but for most organizations it's simply too expensive to implement at the drop of a hat.
Human error can also grind VDI to a halt. All it takes is someone tripping over a wire or installing the wrong software and every virtual desktop in your environment could suffer. Enterprise-wide change management brings technicians, admins and business leaders together to form a review board. The board prevents most accidents by deciding when changes happen, what changes happen and who makes them.
Ultimately with VDI, success comes down to doing what makes sense. Only use VDI if it fits your needs, be sure it can help you now, understand how it works, and test, test, and test again.
15 reasons VDI projects fail
A tail of VDI trial and error
Monitoring tools pave the way for VDI success