everythingpossible - Fotolia

How can IT identify and manage VDI usage fluctuations?

Unpredictable user behavior and boot storms can cause VDI resource usage fluctuations throughout the day. IT can take steps to identify and curtail the effects of these changes.

It's normal to have fluctuations in resource usage in any VDI deployment.

These usage fluctuations often correlate to the load the end users collectively place on their virtual desktops. Although usage fluctuations are common, the underlying infrastructure must be able to handle the fluctuations without diminishing the end-user experience.

What causes VDI usage fluctuations?

Smaller fluctuations in VDI usage tend to be related to unpredictable end-user behaviors. A user might, for instance, decide to play a video or upload a large file, causing a resource usage spike. However, these spikes are usually short-lived and don't cause problems if the virtual desktop density is not excessively high. Overly dense virtual desktop deployments leave few hardware resources available to absorb these usage spikes.

However, a bigger VDI usage issue is the activity spikes that happen due to concurrent user activities -- such as when users first log in and again after lunch. These activity storms eventually subside but can be highly disruptive until they do.

How to spot fluctuations in VDI usage

Identifying VDI usage fluctuations is normally easy to do. Using a virtual desktop daily can give IT professionals a good sense of the ongoing activity patterns.

A list of common VDI processes to monitor, including user login times and session responsiveness.
IT teams should track key VDI processes to avoid a poor end-user experience.

For instance, an IT pro might notice that performance becomes sluggish at 9:00 a.m. for about 10 minutes and again at around noon when everyone heads to lunch. Even though such observations are subjective, IT can quantify sluggish performance during heavy activity using performance monitoring software. To understand why performance is suffering, you will need to run the performance monitoring software on both the virtual desktop and the host server. That way, you can determine if too few hardware resources are allocated to the virtual desktops or if the host server is being depleted of its hardware resources.

4 ways to manage VDI usage fluctuations

If VDI usage fluctuations are causing the virtual desktops to become sluggish with noticeable performance drops, several things can improve the end-user experience.

1. Pre-boot automation

One way that you may be able to overcome some of the more serious usage fluctuations is to take advantage of automation. Performance problems often stem from activity storms resulting from multiple users engaging in heavy activity simultaneously.

One specific type of activity is called a boot storm. A boot storm refers to the activity generated when a collection of virtual desktops power up at the beginning of the day. If boot storms are a problem in your organization, you might consider using IT automation to boot the virtual desktops before the users' workday starts. That way, the boot process won't interfere with the end-user experience.

Even if you do pre-boot the virtual desktops, activity storms are still likely to occur at the beginning of the day as users log in and begin opening applications. It may be possible to further reduce the effects of these storms by configuring some applications to pre-load. However, take care in doing so to avoid consuming excessive hardware resources by loading applications unnecessarily.

2. Workload redistribution

IT pros can also use automation to manage VDI usage fluctuations by redistributing workloads.

IT pros can also use automation to manage VDI usage fluctuations by redistributing workloads.

Suppose, for example, IT knows that a major resource spike occurs daily at noon. Depending on which hardware components are being overutilized, they may be able to use an automation engine to automatically power up an additional server node and then live-migrate some of the virtual desktops to that node just before the resource spike occurs. This way, more hardware is available to cope with the demand.

When the usage spike is over, IT can set the deployment to automatically live-migrate those virtual desktops back to their original location and shut down the extra server node to save on power and cooling costs.

3. Reduce the virtual desktop density

Another option for coping with usage fluctuations and activity spikes is to reduce the virtual desktop density. Virtual desktop density refers to the number of virtual desktops that run on a given host. The greater the density, the higher the return on your server hardware investment, so organizations often seek to achieve the highest possible virtual desktop density.

The problem is that performance generally decreases as the virtual desktop density increases because more virtual desktops compete for a finite set of hardware resources.

If you have problems with virtual desktops performing poorly, it may be worthwhile to decrease the virtual desktop density by adding extra hosts.

4. Host the virtual desktops elsewhere

Another option for overcoming virtual desktop performance problems is migrating your virtual desktops to another platform. Virtual desktop migrations can be a major undertaking, so it's best to use this option only if your existing hardware cannot deliver adequate virtual desktop performance.

If you decide to rehost your virtual desktops, there are two common options. One is to host the virtual desktops on premises using hyper-converged infrastructure (HCI). The advantage of this approach is that HCI deployments use a series of nodes that are identical to one another from a hardware perspective. This allows you to figure out how many virtual desktops can be comfortably hosted on a given node and then extrapolate how many nodes are needed overall. As your virtual desktop needs increase, you can easily add more nodes.

A second option is to host your virtual desktops in the cloud. While you can build a cloud-based VDI deployment from scratch, you could use desktop as a service (DaaS) instead. DaaS generally ensures a certain level of performance and frees you from having to maintain the underlying infrastructure.

Dig Deeper on Virtual and remote desktop strategies

Enterprise Desktop
Cloud Computing
SearchVMware
Close