When a VDI user contacts your help desk because a key application is slow, you can probably expect a bunch more calls just like it. Too often, these callers hear, “Everything looks fine from this end.” Here’s how to offer VDI users a better answer now — and fewer reasons to call in the future.
In a virtual desktop infrastructure (VDI), user desktops run within a virtual machine that lives on a datacenter server.
VDI has been gaining popularity over the last decade because because, as opposed to the traditional client/server model, it enables personalized desktops for each user. At the same time, it allows IT to centralize security and other key functions.
As VDI technology has matured, network managers have mostly learned how to avoid “storms” caused by many VDI users simultaneously booting up, installing scheduled updates, etc. But other productivity-killing application bottlenecks in VDI environments have proven more elusive.
To improve application performance within your VDI, improve your forensic capabilities and overall real-time visibility:
1. Don’t wait for another real-time failure to diagnose the problem
Some application performance issues can go undetected in a VDI environment because users don’t notify IT, or at least not soon enough to diagnose the issue.
One reason for this is that groups of VDI users often aren’t physically in the same place, so there’s no cubicle “prairie dogging,” where folks stand up and take an informal poll: “Hey, is Office 365 down for anybody else? Yeah? Okay, somebody call IT.”
Instead, the first instinct for individual users — especially employees with a non-IT mindset — is probably to go get a coffee or something, and wait for whatever’s going on with the machine to work itself out. And from their perspective, that actually seems to do the trick sometimes.
Multiply that user’s experience by 100, 1,000, or 10,000 more affected employees, however, and the lost productivity can be staggering.
When you see a pattern of poor performance for a mission-critical application, don’t try to wait for another outage to try to diagnose the issue in real time. You need to dig deeper, using a monitoring tool that gives you forensic capabilities.
Netreo recently worked with a company that was having trouble identifying the cause of a performance issue with a critical ERP application — a legacy platform the company had built itself.
We showed them how to access readily available metrics showing which processes were running at the time of day and week that the problem had occurred. It turned out they had a compression job running as part of a backup that was eating all the CPU on their VDI system.
Your network monitoring should have the capability to quickly get to network performance data from hours, days or even months ago.
Waiting for the issue to recur — on the theory that next time you might be actively monitoring your VDI metrics when it happens — is a recipe for repeating a preventable disaster.
2. Set up notifications for new applications added by users
An advantage of VDI is that it can give users more flexibility to customize their machines than they had with the traditional server-based architecture. With VDI, users can have separate OS’s and application packages. But this can backfire.
Does your network management system alert someone when a VDI user installs unauthorized applications (or any new applications, for that matter)? If not, a single user can launch a rogue application that will affect other users sharing the same memory or CPU resources.
This is partly a communication issue. Users need to understand guidelines for installing applications on their own. Assuming some users won’t get the message, however, you need to set alerts for this activity and be sure the right admins are notified to follow up.
3. Use a unified network monitoring system in addition to VDI-specific monitoring tools
The VDI monitoring tools dedicated to one VDI product are essential for setting up, provisioning, and testing of your VDI’s performance.
Once the VDI system is in daily operation, however, you should be using a unified, single-pane-of-glass network monitoring system to troubleshoot potential VDI-related resource contention.
You need the ability to drill down quickly and find out whether, say, a guest is running a heavy job that’s thrashing a disk array other VDIs depend on.You also need to be able to see at a glance whether the hardware a VDI is running on may be the source of the bottleneck.
Once you’ve diagnosed the issue, you may not be able to use the VDI’s dedicated monitoring system to do anything about fixing the issue. Again, that’s usually a job for a unified network monitoring system.
VDI has gained popularity in part because it gives IT centralized control — up to a point. But with centralized control comes the need for centralized visibility into the VDI as a part of your whole system.
You’ll need that visibility to protect VDI users from other VDI users who rely on the same resources.