This blog is part four of the “Making VDI Work” blog series. If you’d like to learn more about how to make VDI work for your organization, check out our previous installments How to Design the Ideal VDI or Hosted Desktop Solution, Things to Consider about Client Devices, and Choosing A Display Protocol
With summer just around the corner, and despite the rainy New England weather, our thoughts are turning to outdoor getaways. Unfortunately, as we all know too well, getting away from the office doesn’t always mean getting away from work. Thankfully, that’s one of the great benefits of building a hosted desktop solution. Your users can roam and still access their desktops and applications, without physically carrying them around.
So, to help you keep your users productive while they sit on a beach with their tablets, we bring you the fourth installment of our “Making VDI Work” series!
First, don’t limit your view to virtual desktops
If you’ve been following along, our series previously covered the client devices used to connect to your hosted resources, and the display protocol used to transport and render the remote display on that client device. Now, we want to focus on what your users are connecting to, the hosted resources at the top-right the following architecture diagram.
We’ve called this the “Making VDI Work” series, but that’s a little bit of a misnomer. You don’t have to, and really shouldn’t, limit yourself to virtual machines when considering what resources to host in your data center or in a cloud.
The key is to consider the tasks your users need to accomplish and the tools they require to accomplish them. If you answer those questions outside of the context of where and how you plan to host your resources, you’ll design a solution that provides the best end-user experience.
Second, your users’ workflows define your hosted resources
Moving your users’ desktops off of their desks and into your data center or cloud offers you an opportunity to redefine how your users work and what they use. Don’t just migrate their current desktops to virtual machines and call it a day. Instead, take the opportunity to consider how you can optimize resource usage so you can minimize the cost associated with your hosted resources.
I have a background in Usability, so I can’t emphasize enough how important it is to take the time to analyze how your users currently work, and look for ways you can improve that process.
Here’s some questions to keep in mind as you do this:
- What tasks are they completing, and can you define tasks that are common across your user base?
- How compute-intensive are their tasks? Are they watching videos, doing 3D rendering, using a word processor, or performing other tasks?
- What applications are they using, and do they have applications on their desktops that are underutilized?
- What data are they accessing, and where is it located?
- What type of personal devices do they have?
Gather as much information as possible into how users want and need to work, then decide which workflows you want to support.
Finally, what are your options?
Depending on how you answer those, and other, questions, here are some resources you may consider hosting.
- Full virtual desktops: This is standard VDI and gives users access to individual desktop operating systems. Even in traditional VDI, there are different workflows you can satisfy. You can give some users a dedicated, persistent desktop, as if you moved their workstation into the data center. Then, for other users, you can provide a non-persistent desktop, i.e., one that is refreshed or deleted after the initial user logs out.
- Shared sessions: The most common example of a shared session is Microsoft Remote Desktop Sessions, or RDS. In this case, multiple users log into sessions hosted on a single server operating system, and the sessions share the underlying operating system and installed applications. For task workers, particularly for those that need access only to a particular application, shared sessions are a good solution. When thinking about providing sessions, don’t limit yourself to Microsoft Windows operating systems. Linux operating systems are inherently multi-user, so you can potentially save money by switching to Linux.
- Physical systems: For users or applications that require significant compute resources, consider switching from workstations under their desk to rack mounted blades or workstations in a data center, or to an HPE Moonshot System. Often, hosted physical systems can be shared by multiple users, increasing the utilization of expensive resources. Also, many applications access data that is also hosted in the data center, so collocating the compute near the data may improve application performance.
- GPU-enabled virtual machines: If providing individual blades or workstations seems too expensive, consider GPU-enabled virtual machines. Using virtualization with workstations that have multiple GPUs can often provide the performance necessary for graphics-intense applications, while requiring less hardware.
- Applications and containers: Often, it’s not the operating system that users care about or need access to. It’s the application. Moving some workloads into containers can increase the density of your hosted solution.
Key take aways
As always, for the skimmers among you, here are a few key takeaways when considering what to host in your data center
- Don’t design a solution that works only for a single user group. Look at all of the workflows in your organization where hosting a required resource can increase productivity and/or save money, so you can design a solution that satisfies all of those needs.
- Persistent virtual machines are not your only options. Consider a mix of virtual machines, RDS, hosted physical systems, shared applications, or maybe even containers.
- Consider if switching to Linux is feasible and cost effective for you. If users require access only to an application, and won’t be interacting with the operating system, switching from Windows to Linux can save money.
Most importantly, you should think about what you want to host without thinking about where and how you want to host it. If you wedge yourself into a virtualization stack or public cloud solution too early, you may miss the opportunity to host a wider array of resources, and the opportunity to give even more users the chance to work from the beach!
In our next installment, we’ll finally talk about where to put your hosted resources. Until then, where’s my Mai Tai?