Just as you wouldn’t buy the same desktop or laptop for a contact center worker as you would for an architect running AutoCad, virtual desktops are also not “one size fits all.” Instead, they offer an opportunity to right size the desktop to each use case, ensuing end users have the resources they need without over provisioning and overspending. Virtual desktops also have the added benefit of more easily adjusting resources, such as RAM, as business needs change or to support new application requirements – without purchasing and deploying new hardware.
When sizing virtual desktops, the same components are under consideration as with physical desktops. The main components are RAM, CPU and Disk Size (capacity). vGPU or accelerated graphics capabilities are also becoming more mainstream. Here’s a look at how each component differs within virtual desktops from more traditional, physical deployments and some of the most common sizing options for typical virtual desktop use cases.
Not sure what you need? Talk to a Dizzion Solution Engineer!
RAM is one of the biggest components that drives the cost of the virtual desktop as it is a big part of the overall cost of the physical hardware necessary to run a virtual desktop environment and can also affect licensing costs.
One thing to consider when sizing a virtual desktop is that you may not need the same amount of RAM that a physical desktop may require. Simply taking the amount of RAM a physical desktop is using might not be a good benchmark for how much a virtual desktop needs because Windows has background processes that can consume however much RAM is available to it.
Also, in a VDI environment, engineers optimize the operating system image. This means disabling all non-essential processes and services. Great care is given to having an efficient OS image that runs in a small footprint.
In a virtual environment, hypervisors can further reduce the need for RAM. Dizzion uses VMware to power our underlying infrastructure, which has features in ESXi that can scan a virtual machines memory. After all virtual machines are scanned, ESXi can either de-duplicate identical pages of memory within a virtual machine (Transparent Page Sharing – TPS) or compress the pages in memory, thus reducing the need to swap memory to disk storage which can cause performance degradation as disk is used for functions memory would typically handle.
Conversely, insufficient RAM impacts virtual desktop performance by causing Windows to perform excessive paging or swapping, negatively affecting performance. In either scenario, all of today’s hypervisors have effective ways to help manage RAM allocations by constantly re-balancing physical RAM capacity and the needs of the virtual machines.
As with any cloud environment, two factors come into play with CPU allocation. The first is the age of the CPU. With each new release, CPU core count and processing power is increased. Be sure you know that your VDI provider has the ability to upgrade/update their cloud environment so you are not stuck using ten-year-old hardware. As your business constantly adds new applications, such as O365, your resource needs may change. To make matters even more complicated, some large providers present processing power into their own proprietary units, which makes it impossible to compare to standard industry metrics such as CPU clock speed (megahertz/gigahertz). Choose a VDI platform that can keep pace with your evolving needs.
The second consideration is how heavily is the CPU oversubscribed? CPU is easily oversubscribed as many virtual desktops can utilize the same physical CPU core. Some providers take too much advantage of this (at the risk of performance degradation) if it creates additional margin/revenue for the service provider. Select a provider that is transparent with their over-subscription policies, or, even better, select a partner that has the ability to deploy a private cloud node dedicated to your use, with your specific over-subscription policies enforced.
Most of today’s VDI use cases do not call for virtual GPUs. Today’s VDI, if properly designed, can handle voice, video and even 3D graphics. However, there are several market factors that are giving vGPUs a lot of attention.
In the past, when implementing a company-wide VDI initiative, one group that had intensive graphics needs would necessitate the need for physical devices for that group. But technology changes quickly, and this is no exception. Today, these intensive graphics needs can be met without sacrificing performance. In fact, companies can mix in vGPU enabled desktops for those who need to view or create 3D renders/models, enabling company-wide deployments across departments and use cases.
Windows 10 users running common applications such as Skype for Business (which is a known resource hog) are beginning to drive vGPU use and it will likely become the defacto standard; but today it largely remains a feature designated for architects and engineers, mainly because the cost does not typically justify rolling out vGPU support to all virtual desktop users.
Generally it is best practices to provide enough disk space for the operating system while user data is best allocated to a different virtual disk volume. This allows IT departments to manage the virtual desktop independent of the users’ data. For example, if there is an issue with the virtual desktop, IT can quickly recreate the desktop from a known good image and then attach the user’s persistent data volume, versus spending hours troubleshooting a stubborn printer issue.
Storage solutions include storage native to the virtual desktop, mapped network drives, or redirected to a customer’s existing storage solution. If customers have existing life (or warranty) left on their existing storage platform, Dizzion can take advantage of that. When it becomes End-of-Life( EOL), customers can avoid heavy investment costs and migrate to a Dizzion hosted/managed storage solution. Dizzion offers the flexibility of either integrating with customers’ existing storage, or providing storage services that are physically next to the virtual desktop infrastructure.
Virtual desktop environments are not one-size-fits all. The applications that a specific use case calls for will determine the most appropriate build configuration. Ensure your solution supports today’s needs and can also grow and scale with your users to meet the needs of tomorrow. To get things started, Dizzion has developed three pre-packaged solutions of our most common configurations and also has a package that allows for a more customized configuration.
Prime Virtual Desktops
Pro Virtual Desktops
Power
Power+ (Custom Configuration)
Regardless of the package selected, Dizzion optimizes the virtual desktop deployment for each organization’s needs. Each virtual desktop group has access to the exact data and applications required, ensuring a secure, high-quality and productive user experience. Each cloud desktop is configured for performance according to vCPU, RAM and disk/IOPS just as you would design a laptop to meet performance requirements. Organizations can also subscribe to differing numbers of Prime, Pro, Power or Power+ desktop configurations to meet their needs. Dizzion experts work with you to identify your use cases and design the ideal virtual desktop environment.
Feb 23, 2023
Chromebooks are an inexpensive end point device that are an excellent option for virtual desktop environments. ChromeOS enables IT to deliver secure devices, fast deployment, and cloud-first management. READ MORE
Apr 13, 2021
Dizzion extends AnyCloud global delivery with new Managed Desktops on IBM Cloudâ, featuring VMwareâ Horizon technology Earlier today, Dizzion announced ...READ MORE
Nov 15, 2018
As the year comes to an end, spend time auditing your virtual desktops and planning for the future so you can have a smooth, productive 2019. READ MORE