A virtual desktop infrastructure (VDI) environment is quickly becoming the choice for many organizations to manage their ever-growing number of desktops. One of the largest benefits of adopting a VDI environment is the drastically reduced strain on operations to maintain workstations. By centralizing user desktops to one location, IT staff spends less time pushing updates or refreshing workstations.
Despite these advantages, VDI doesn’t come without its challenges. While most user desktops have very modest performance demands on their own, together they can present strain on the host. An important task for VDI to be a viable solution is providing a consistent experience across all desktops. However, VDI can easily overwhelm the hardware it’s on, causing latency.
In a typical larger organization, it’s possible that the bulk of employees arrive at eight or nine in the morning and log into their workstations. An overwhelming number of users calling up their desktops at the same time can cause a bottleneck called a boot storm and can delay users’ start up, sometimes even five to 10 minutes each.
If hardware isn’t up to the task to handle this flood of requests, virtualization software can sometimes compensate for some of it. For example, VMware has a feature that will boot up desktops at predetermined times in order to stagger the requests coming into the system.
Amount of Write
Thin provisioning is a favorite feature for those deploying virtual environments on limited space (or limited budget). The downside with this is that increasing storage usage can start to degrade performance. What may have worked great when users were first deployed may start to have serious performance issues when storage approaches higher utilization, even without additional users. The inconsistent performance curve of legacy storage systems can also cause VDI systems that tested fine to fail in production.
Additionally, if a VDI environment has thin provisioning turned on at the hypervisor layer, it can be particularly inefficient at managing write-allocation issues.
Legacy storage is often the culprit of VDI performance problems. While VDI can make use of whatever storage an organization currently has, it may not perform as well. In fact, many organizations report degraded performance after switching to VDI.
A tiering/caching approach — which is comprised of a combination of higher-performance solid state drives (SSDs) and hard disk drives — is a good solution to this problem. The system can then automatically provision blocks of data between the two types of storage. For example, the frequently requested reads like boot images would be kept on the SSD, which would satisfy the boot storm demand placed on the system. This will make end users happy because their VDI desktops will perform well.