5 Misconceptions About Optimizing WAN Traffic For Virtual Desktop Infrastructure

More and more enterprises are turning to virtual desktop infrastructure (VDI) as a means to reduce the expenses normally associated with traditional endpoints, such as PCs, laptops and tablets. However, successful implementations seem to be far and few between, with many VDI projects never leaving the pilot program stage.

About a year ago, Gartner estimated that there will be as many as 20 million virtual desktopsin place by 2014. A few months later, CDW found that 90% of businesses are considering or implementing client virtualization projects, most of them within the next 12 to 24 months. However, CDW also reported that companies were finding that client virtualization is more complex to implement than they realized, that ROI is difficult to calculate and that training end users can be a challenge.

Almost every VDI deployment conversation includes a critical question: How well will VDI desktops perform for end users? Add in remote workers and WAN connections, and the performance conversation becomes all-encompassing. Simply put, VDI has to perform as well as, if not better than, whatever endpoint technology is being replaced.

Therein lies the dilemma: How can an administrator improve the performance of VDI--a technology that is normally preconfigured for maximum performance--without harming the chances for deployment? Perhaps the best answer lies in what not to do--in other words, mistakes to avoid when scoping out ways to enhance the end user experience with VDI-based endpoints.

Some of those mistakes border on the obvious, while others seem unique to the intricacies of VDI. However, they all share the same result--failure to correct or anticipate performance problems.

One of the first mistakes that adventurous network managers make is to use the ideology of "accelerate all." That is where an application/WAN acceleration technology is put into place to compress and accelerate all traffic over the wire. The primary problem with that ideology is that the "bad" is often accelerated with the "good." That means low-priority traffic, such as emails or file sharing, gets boosted--often at the expense of critical traffic such as VoIP and VDI display protocols. The best way to approach that situation is to accelerate a few types of traffic while ignoring the rest.

Another mistake that network managers can make with VDI is the quick-fix solution--a situation similar to above, where acceleration technology is put into place without looking at either the big picture or the types of traffic on the network or future application loads. In many cases, a quick fix is just that, a quick fix, and more serious performance problems rear their ugly heads over time. The trick here is to plan performance management using a unified approach. In other words, select a unified performance management product instead of a one-trick-pony quick fix.

Many network managers often get tripped up by another issue--complexity. It is critical to use the simplest methodology for increasing performance; otherwise, acceleration technology can become too unwieldy to manage, measure and troubleshoot. That is a situation that can be solved using the age-old KISS (Keep It Simple and Short) principle.

Some mistakes don't become evident until it is too late or aren't thought of as mistakes until there are repercussions. One of those mistakes falls under the axiom of "expect the best but plan for the worst." A case in point is bypass testing (more commonly thought of as failover). Often overlooked, yet extremely critical, is what happens if acceleration technology fails. Simple questions such as "Will traffic still flow during a failure?" or "How does traffic route around a failed acceleration device?" are key when thinking about traffic interruption due to failures. Network managers need to implement equipment that is fault-tolerant but also test that fault-tolerance. The worst time to find out it doesn't work is during an actual failure.

Another common mistake network managers make when dealing with WAN and VDI acceleration technologies falls under the heading of misconfiguration. Many of the acceleration products on the market are resilient enough to still work, albeit poorly, when incorrectly configured. Many network managers will find that there are a surprisingly high number of incorrect configuration settings possible on networking devices concerning Ethernet negotiation. Some are as simple as making sure auto-negotiation is turned on both sides; others border on the more esoteric, such as one side of a connection being set to half-duplex while the other side is set to full-duplex. The lesson here is to perform due diligence when configuring the various networking components that make up the link between a VDI host and an endpoint.

Learn more about Strategy: Predictive Analytics for IT by subscribing to Network Computing Pro Reports (free, registration required).