Many VDI shops simply measure infrastructure performance and assume a healthy infrastructure means a healthy VDI user experience. Unfortunately, the two are less tightly linked than it might appear on the surface and using infrastructure load to predict user experience is a mistake.
Instead, IT administrators' VDI experience monitoring approach must examine what their users' application experience is actually like and measure what the applications feel like for users. After all the whole point of VDI is to deliver applications to users fast enough to provide a good user experience. Access to a slow application is only a little better than not having access at all.
A poor VDI user experience is often the result of a single resource or a single misconfiguration. A user might have a drive mapping to a retired server, a WAN link may be heavily loaded or data might be coming from a remote data center. None of these things make a VMware vSphere cluster look busy, for example, but any one can cause a poor user experience.
On the flip side, a heavily loaded infrastructure can provide a good user experience. In the service provider world for example, systems that run at 90% utilization can still deliver a good user experience.
How can admins tell what's happening with VDI user experience?
Part of the challenge of identifying problems is that users simply do not have the vocabulary to tell admins what their experiences are like. The classic, "everything is slow" complaint is usually not specific enough to be helpful unless admins already know something is wrong on the back end.
The performance problems could be the result of an overloaded VM, a slow WAN link, or an office visitor who left his BitTorrent software running and is making the network slow for the whole site. What admins really need for effective VDI performance monitoring is a specific report that lets them know Microsoft Word is loading fast but is slow to open three different documents on the sales share, for example. They need a way to measure their users' experience and gain insight into the cause of any issues.
How to deal with the unpredictability of users
Measuring real users is hard, they simply act so randomly. To compare the user experience from day to day or from data center to data center, admins must remove the randomness. There are a couple ways to measure user experience:
Average a whole population: A large enough group of users behaves quite similarly over time. The peculiarities of each user's individual behaviors balance out when admins measure a large group. This method gives admins an overall measure of users' experience accessing their VDI deployments.
Measure synthetic users that behave in predictable ways: This VDI performance monitoring method involves creating some automated application activities to mimic real users. Admins initiate desktop sessions that launch these automated applications and measure the user experience. The experience for a consistent set of application tasks lets them infer the VDI user experience for real users. A crucial part of synthetic users is that the automated tasks must closely resemble real users for the measurement to be useful. Admins need a tool that allows them to easily add new applications and tasks in those applications.
Ready to put a VDI project plan in place? Take this quiz to prove it
VDI projects can fail easily. See how well you're prepared for VDI with this quiz, which asks about how to track user experience, what makes a successful pilot program and more.
Once admins measure the overall VDI user experience they know whether they have a general problem. If they have a poor VDI user experience they will want to know where it is poor and which application or login phase is slow. This is where detailed logging of the user experience monitoring is crucial. If admins measure overall good user experience but a single user reports a problem then they are in a better position to troubleshoot the specific problem.
There are a few VDI monitoring tools that measure user experience. Admins' existing VDI management tools may include the capability, but user monitoring is central to Liquidware Labs Stratusphere UX, as well as to Goliath Technologies' products. Synthetic user monitoring is the only function Login VSI's Login PI product performs but is just one feature for ControlUp.
The user experience of accessing applications is the ultimate test of whether a VDI deployment is doing its job. Measuring and tracking the VDI user experience should be a foundation to supporting any VDI deployment.
What can VDI monitoring tools really do?
Find a middle ground between UX and security
How important is mobile VDI user experience?