Desktop virtualization technology has surged in demand recently, being requested by customers across all industries. But deploying a virtual desktop infrastructure (VDI) for your customer is not an easy task if you don't fully understand the technology. In this FAQ, Rick Brown, director of managed services for Alteritech Inc., answers some of the most frequently asked questions about virtual desktop infrastructure services, including questions about cost, thin clients and more. In the podcast, Rick discusses how to market virtual desktop infrastructure services to customers and explains which features to highlight when describing the benefits of VDI.
• What are the cost benefits of a virtual desktop infrastructure (VDI) deployment?
• Can all of a company's desktops and laptops be virtualized?
• Should thin-client hardware replace the existing desktops in a VDI?
• How long should a virtual desktop infrastructure solution deployment take?
• What is the best way to address network performance issues with VDI?
• How should the VDI architecture be designed for resource load?
• What is the best way to determine virtual desktop infrastructure host server sizing?
• When it comes to each VDI user, what's better: virtual machine pooling or dedicated VMs?
What are the cost benefits of a virtual desktop infrastructure (VDI) deployment?
VDI is well-suited to address the unique challenges of managing an enterprise desktop environment. With traditional desktop environments, it can be very difficult to adequately manage issues such as data security risks, patch management, disaster recovery, regulatory compliance requirements, high availability needs, on-demand scalability and end-user performance expectations. VDI can help address all of these issues due to its highly available and centralized back-end infrastructure and associated technologies. However, while VDI may provide some long-term ROI benefits, those looking for an immediate reduction in desktop infrastructure costs will likely be disappointed -- the up-front infrastructure and deployment costs can be substantial, especially when implementing VDI in-house. Outsourcing VDI to a service provider can help to minimize these up-front costs.
Can all of a company's desktops and laptops be virtualized?
In all likelihood, the answer will be "no" -- but it depends on the user demographics within the organization. Knowledge workers who primarily use business productivity applications are the prime candidates for migration to a virtual desktop infrastructure solution, while mobile workers with laptops are almost never candidates for VDI. Additionally, those workers with high-performance desktop needs -- such as engineers or creative professionals -- can seldom be successfully integrated into a VDI architecture, due to performance issues with applications such as photo editing and 3D graphics. A VDI feasibility assessment is typically the first step taken to analyze and categorize the desktop environment in the organization, and the results can then be used to design a pilot VDI project.
Should thin-client hardware replace the existing desktops in a VDI?
Not necessarily. True thin-client terminals provide additional benefits such as reduced IT support and hardware costs. But in most cases, the existing desktop PC hardware can easily be converted for use as a remote desktop, either by installing a specialized thin-client OS (Linux) on it or customizing Windows by uninstalling everything but the remote-access client and launching it upon startup. A good approach is to use your existing desktop hardware as remote desktop clients and phase in true thin-client hardware over time as the desktops are retired.
How long should a virtual desktop infrastructure solution deployment take?
Most in-house VDI architectures can take anywhere from four to 12 months to assess, analyze, architect and deploy. A good portion of this time is spent assessing the existing desktop environment and designing a suitable VDI replacement. A pilot implementation will usually last two to six months to ensure the design goals can be met in a full-scale implementation. VDI solutions from a solution provider bring the benefit of turn-key solutions where resources can be allocated on-demand in most cases. This can significantly reduce the implementation time to as little as several weeks for a pilot project.
What is the best way to address network performance issues with VDI?
Ensuring that the end-user experience is as close to a real desktop as possible is one of the challenges that must be overcome with a virtual desktop infrastructure solution, particularly if the back-end VDI server infrastructure is located in a remote data center. Each VDI session must have adequate bandwidth to ensure that end-user performance expectations are met. When session traffic traverses a LAN, this is seldom an issue. However, over high-latency WAN links this can become problematic, especially over the Internet. This can be addressed using dedicated VDI WAN links to the data center, traffic management technologies and/or remote access protocols better suited for latent connections.
How should the VDI architecture be designed for resource load?
Ensuring that adequate resources are designed into a new virtual desktop infrastructure design is one of the most critical factors for success. The key to doing so is assessing the existing desktop CPU, storage, networking and memory usage patterns in the organization. Metrics should be monitored and collected over a sufficient timeframe (two to four weeks) from representative samplings of the desktop population to form a performance baseline for each category of user class. In addition, every application should be tested within the VDI environment to ensure it performs adequately and correctly.
What is the best way to determine virtual desktop infrastructure host server sizing?
This is a very difficult question to answer without a good baseline assessment to draw upon. But in general, you can estimate approximately five to seven virtual desktops per CPU core, assuming a relatively modern CPU such as the Intel Xeon 3GHz quad-core processor. So a single server with dual quad-core processors could accommodate a workload of approximately 40 to 50 virtual desktops, assuming adequate memory exists on the host.
When it comes to each VDI user, what's better: virtual machine pooling or dedicated VMs?
Virtual machine (VM) pooling refers to maintaining a pool of available VDI VMs that are always available to be "checked out" by users. A user may get any VM within the pool and will not get the same VM each time they log into their desktop environment. By contrast, with dedicated virtual machines, the user will always get the same VM each and every time they log into their desktop environment. Dedicated VM environments are typically easier to set up and do not require the services of a "connection broker" like VM pooling requires. The connection broker provides the pooling services and manages the provisioning and tear-down of VMs based on organizational policy. Most organizations will prefer to use a connection broker due to the advanced features that it provides, such as automatic provisioning and the deployment of VMs. So in most cases, VM pooling will be preferred over dedicated VMs.
About the author
Rick Brown is director of managed services for Alteritech Inc., an IT infrastructure consulting and managed services firm focused on providing technology solutions. With 17 years of IT experience including strategic planning, architecture design, implementation and operations of complex, distributed environments, he now focuses heavily on virtualization, storage and data center optimization. Prior to joining Alteritech, Rick was a senior manager at Acumen Solutions, a leading business and technology consulting firm with offices across the United States and in Europe.