Sergey Nivens - Fotolia
There are many third-party software tools and products that enable data center managers to plan effective virtual desktop infrastructure deployments. From freeware monitoring to internal testing methods, a range of preventive measures can help organizations avoid VDI performance issues and end-user discomfort, as well as free up storage and staff resources.
VDI planning and implementation
Ideally, a virtual desktop infrastructure (VDI) should be thoroughly tested and benchmarked prior to going live. Basic freeware tools like ioMeter can be used to do some initial measurements; however, to get a more in-depth picture of how the VDI environment will perform under production loads, more advanced tools like Login VSI can be used to simulate close to real-world end-user desktop activity.
Likewise, you can leverage testing appliances, like those from SwiftTest, to accurately simulate how well the underlying VDI storage infrastructure will perform based on a variety of real-world testing scenarios. The insights gained from using benchmark performance tools will enable you to identify up front which components will be needed to support data movement on the network infrastructure side as well as in back-end data stores.
What's more, products like Login VSI and SwiftTest can be used post-deployment to help you gain visibility into any potential performance bottlenecks that may creep into the environment as you plan for changes to the infrastructure, such as adding more virtual desktops to another network segment.
There are also VDI assessment tools, such as those from Liquidware Labs. That offering relies on software to perform an assessment of the environment and then suggests which users will be most suitable to having the desktop experience virtualized and which ones will not. These products are unique because they calculate the workload required by the current desktop environment, rather than the potential environment.
From a day-to-day management perspective, some industry experts suggest monitoring storage IOPS as a way to judge system loads. This is fine for establishing a general baseline of VDI performance during peak load conditions; however, it's not an ideal tool for troubleshooting problems in real time. The challenge here is that unless you happen to observe a system bottleneck when it occurs in real time, it will be very difficult to determine whether the performance issue is storage or network related; and even then, it will be difficult to ascertain the root cause.
By augmenting existing data storage resources with products designed to accelerate VDI performance, IT organizations can reduce the initial capital outlay required to seed a VDI deployment.
While there are many companies that offer virtual server monitoring solutions, it's important to distinguish VDI from virtual machine (VM) monitoring. For this reason, companies like eG Innovations, SolarWinds and Xangati may provide a more comprehensive suite of VDI performance monitoring technology. Some universal VDI performance issues, like boot storms, aren't nearly as prevalent in a virtual server infrastructure. The aforementioned companies' products provide in-depth analysis and monitoring of infrastructure-related conditions unique to VDI environments.
Xangati provides a VDI performance monitoring framework with DVR rollback capabilities to review both historical and real-time I/O activity. This is particularly useful for isolating storage performance issues and determining the root causes of activities that cause contention storms. Likewise, eG Innovations has management tools that take historical data from the VDI infrastructure and establish a baseline for what constitutes normal performance activity. When baseline activity is breached, VDI administrators are notified.
For example, in a large university campus setting, VDI administrators could plan for peak demand activity at the end of a semester as students use their virtual desktops for exam preparation, term paper submissions and so on. On the other hand, a rush of students accessing a recorded lecture the day before an exam may be less predictable. Having an automated system to track historical activity can at least provide VDI administrators with some insight as to how to proactively plan and react to such events.
Managing VDI storage tiering and capacity usage
Many organizations are leveraging networked-attached storage (NAS) devices to serve up VDI storage resources. Oftentimes, these are existing storage assets that are re-purposed. NAS devices work very well for servicing VDI workloads provided their storage capacity and performance is carefully monitored and managed.
The inherent challenge is that many NAS systems are specifically optimized for write I/O performance. As a NAS storage back end approaches 40% utilization or higher, read I/O begins to suffer as write I/O workloads are queued ahead of read I/O activity. This spells trouble for VDI deployments because the majority of VDI workloads are read I/O-oriented. To circumvent this limitation, companies like Alacritech and Avere Systems have developed NAS caching appliances to "front end" existing NAS storage systems.
By providing an intelligent front-end caching appliance, read I/O workloads can be automatically loaded into flash or solid-state drive storage to provide highly optimized read I/O performance. This has the dual benefit of extending the life of the existing NAS array, as it can be dedicated to serving up write I/O activity -- which it was designed for -- while also acting as a relatively low-cost storage repository for inactive data sets.
Once VDI storage capacity has been provisioned to optimize both read and write I/O activity, you can address the question of the best way to monitor ongoing storage capacity. The onboard storage management tools from the NAS caching device and the back-end NAS array, for example, can be used to determine growth rates for capacity planning purposes. Leveraging a VDI monitoring framework to automate the collection of this data is particularly useful, especially for VDI deployments in excess of 1,000 desktops.
Overall storage growth is rarely, if ever, proportional across the various users consuming VDI storage resources. For this reason, it helps to gain better insight by isolating the biggest consumers of VDI storage resources. VDI monitoring frameworks, like those listed above, provide tools that automate the collection of storage usage data across wide VDI user deployments to enable more accurate storage capacity planning.
Protecting VDI data
As with many technology projects, data protection or backup is often treated as an afterthought. This is a pitfall that IT planners will want to sidestep from the outset because backing up a VDI environment may impact network bandwidth and result in the consumption of SAN storage resources. Virtual desktops consist of either static desktop images, with some minor user customized settings such as those deployed for shift workers in a customer call center, or remote desktops that are frequently disconnected from the corporate network for knowledge-based workers such as sales reps working in their respective territories.
In either case, automating the backup process so end users aren't charged with protecting their own data is the most consistent and reliable method to ensure that critical business data assets are safeguarded.
In the case of a remote end user, there are a variety of backup tools that nonintrusively protect data each time an endpoint device is connected to the company network.
Through pervasive backup technologies such as change block backup, compression and deduplication, the data can be efficiently trickled over the network for safekeeping at a centralized data center facility -- whether it is on the premises or at a third-party cloud backup provider site. For static virtual desktop environments, data can be centrally managed on a NAS or SAN device and be periodically protected via disk snapshots that are offloaded to a disk backup appliance system and/or tape library.
From a data recovery standpoint, backup tools that provide self-service restore capabilities for individual documents or files will contribute to end-user VDI satisfaction while lifting the burden of fulfilling these types of mundane requests from IT help desks or administrators. In the event of a lost or damaged laptop in the field, IT administrators can manage the data restoration process centrally and expedite shipment of the new device as policies dictate.
End goal: Accelerating VDI performance
For businesses that support a large desktop end-user base, VDI enables better protection of corporate data while relieving IT from the continual churn of responding to desktop crashes, viruses and so on. What's more, greater IT efficiencies and savings can be derived by consolidating desktop data into a centralized storage and backup repository.
Unlike server virtualization, however, VDI hasn't enjoyed the same level of relative adoption. This is due, in large part, to concerns over the end-user experience and the total cost of ownership to deploy and manage VDI compared to a traditional desktop infrastructure.
By augmenting existing data storage resources with products designed to accelerate VDI performance, IT organizations can reduce the initial capital outlay required to seed a VDI deployment. As importantly, end-user quality of service can be more reliably managed by leveraging third-party software tools that collect and aggregate performance-related data for enhanced centralized command and control.
About the author:
Colm Keegan is an analyst at Texas-based firm Storage Switzerland LLC. He has been in the IT industry for 22 years. His focus is on enterprise storage, backup and disaster recovery.