ptnphotof - Fotolia
For a VDI deployment to be successful, the end-user experience must deliver performance that can at least match that of a physical desktop. The best way to ensure end-user satisfaction is to use endpoint monitoring to verify the VDI deployment's performance is up to snuff.
VDI monitoring really isn't too different from server performance monitoring, which typically involves tracking performance metrics such as CPU and memory utilization, network bandwidth consumption, and the number of read and write operations.
Despite the similarities, two main factors make VDI monitoring more complex than server monitoring: scalability and finding meaning in metrics.
Why is VDI monitoring complicated?
Monitoring a server involves watching the performance of a single physical or virtual machine. Conversely, VDI almost always consists of multiple servers working together. As such, one of the big VDI monitoring challenges is scalability: Admins must simultaneously monitor the performance of each of the servers that plays a role in VDI.
The second factor that complicates VDI monitoring is translating the raw performance metrics into something meaningful. For decades Microsoft has, for example, given its Windows Server customers guidance on performance monitoring counter threshold values. If, for instance, a Windows Server's CPU utilization remains above 80% for an extended period of time, then it may indicate a CPU bottleneck -- although other factors can artificially increase CPU utilization.
Now, imagine a VDI server clocks in with an excessively high CPU utilization value. What does it mean for the end-user experience? It is really difficult to say for sure. The user might suffer severe performance problems as a result of insufficient CPU resources. On the other hand, depending on where the high CPU utilization occurs within the VDI architecture, the end user might be completely unaffected. The point is, effective VDI endpoint monitoring requires a way to map resource consumption to the end-user experience and vice versa.
How to make sure VDI monitoring is effective
In some cases, admins can apply standard server-level performance monitoring techniques to VDI deployments. This is especially true for very small VDI deployments. As the VDI deployment size increases, however, it becomes increasingly impractical to rely on standard performance monitoring techniques. It is far more effective to adopt a monitoring application specifically designed for VDI.
VDI monitoring tools vary widely in their scope of operation, but one thing they have in common is that they attempt to quantify the end-user experience. For example, some vendors benchmark known processes to look for anomalies.
Take the user login process. Users probably log in to virtual desktops several times a day, every day. As such, it is easy for VDI monitoring software to determine how long the login process should take. If the average login duration is 10 seconds, and logins start taking 20 seconds, there might be a problem. At the very least, the monitoring software would send the administrator an alert. Often, however, the monitoring software examines the individual VDI components to determine the root cause of the problem.
One way VDI monitoring software can determine the underlying cause of performance problems is to look at the similarities and differences between user sessions. Suppose again the login duration should be 10 seconds, and some users experience 10 second logins while others trudge through 20 second logins. The VDI monitoring software might trace the sessions to determine which server-level resources users are working with. In doing so, the software might determine all the virtual desktops experiencing slow logins are hosted on the same server. The software could then compare that server's processes and resource utilization against those of VDI hosts performing properly. As a result, the software could quickly determine the cause of the problem, or at least point the administrator in the right direction.
VDI performance monitoring tends to be more complex than standard server performance monitoring. Several vendors, including Citrix and VMware, offer VDI-specific monitoring tools to help administrators quantify the end-user experience and quickly respond to problematic conditions.
How to avoid VDI performance problems
Solid-state storage can boost VDI performance
Take VDI demand spikes head on