Executive Q&A: NVIDIA’s Chip Carr on Evolving the Telework Environment

MeriTalk recently sat down with Chip Carr, Senior Manager of Virtual GPU Technology at NVIDIA to discuss how Federal agencies have transitioned to telework environments while introducing new technologies, applications and collaboration tools – and what Carr sees going forward.

MeriTalk: What advice have you given agencies that need to expand or improve telework capabilities over the last few months?

Carr: The first advice NVIDIA gives to agencies is that they do not need to limit their VDI deployments to non-graphics users, because just about any application can be virtualized and just about any user can use a virtual machine (VM) and have a great experience working remotely by leveraging NVIDIA virtual GPU (vGPU) technology. Even though Federal leaders initially focused on VDI for providing basic office functionality, much more can be offered thanks to GPU-accelerated VMs, including access to video collaboration tools, high definition monitors – up to 8K – and multiple monitors versus just one. And, you can support higher-end use cases as well for engineers, analysts, and data scientists.

Desktops for these users have traditionally been very difficult to virtualize because they need GPUs. Once we introduced NVIDIA vGPU technology in 2013, these end users could now access their applications and data remotely, and we’ve continued to expand this functionality ever since.

MeriTalk: What do agencies need to do to expand or support telework moving forward?

Carr: There are two areas that agencies should focus on to prepare for the future of telework. First, employees are spending more time using video and collaboration tools, such as Webex and Zoom, and must have a good user experience to remain productive. NVIDIA testing has shown that running these applications in a CPU-only VDI environment can result in a poor experience which is almost unusable. Agencies can prepare themselves for the future – as applications become even more robust – by deploying a powerful VDI environment now. Second, agencies must expand support for engineers, analysts, and data scientists to also leverage the VDI environment, because with NVIDIA vGPU they can access applications in virtual workstations that are indistinguishable from their physical workstations.


MeriTalk: How have you worked through some of these challenges with agencies during this time as they try to figure out remote work?

Carr: When we engage with a customer, NVIDIA always starts by understanding the problem so that we are in the best position to help solve the challenges. We take the time to understand the employee workflows and the types of data they use. And we work closely with the software development community, for example ESRI, to ensure our combined solution works well together – which in turn helps our customers. As we start to understand the needs and consider the variables, we get a picture of the user profile. The team can then create an optimal user experience.

Videoconferencing is a key use case. Everyone working from home is using collaboration tools – Zoom, Skype, Microsoft Teams, WebEx. Seeing one another on camera is helpful – it helps us psychologically, it helps us connect, it helps us do our work. Those applications all benefit from GPU acceleration. A virtual desktop without GPU is going to struggle to provide a quality user experience because you need the GPU to handle the video stream. That is where vGPU comes in, delivering the user experience and performance employees expect.

When we start to talk about knowledge workers, you have to think about everything users are doing all day, and what they need. Good communication is critical to all functions of the mission. Users get bad video and choppy audio when the virtual machine is overloaded and can’t process everything fast enough. Adding GPUs to the virtual desktop significantly improves video and audio performance because you free up the CPU resources. Other use cases include virtual workstations running applications like ArcGIS Pro or CATIA, for example. These applications would be pretty much unusable in a virtualized environment without a GPU. As we work through challenges with customers – it all comes back to key questions: What are you doing? How do you do it? What apps do you use? Tell me about your data.

MeriTalk: What type of agencies have had success making the adjustment to full telework, or are close to it? Can you share some use case examples?

Carr: We have seen many examples of agencies who have made a successful adjustment to telework. Some were well positioned in advance, and leveraged vGPU technology early on, and some quickly adopted these strategies once the pandemic hit. This is across the DoD, civilian agencies, and Intelligence Community as well.

One of the biggest barriers for VDI is a poor user experience. Once people feel that they can’t be productive working in a virtual machine, they don’t want to use it. The types of agencies and departments that have had success are the ones that offer VDI with the user experience in mind. Providing desktop performance remotely without sacrifices, and making sure people can work efficiently and productively without frustrating delays or dropped frames can be a key to success.

Those with virtual desktops, especially VDI with vGPU, didn’t miss a beat. Those with VDI just for use within their office environment tried to connect with a VPN, and had challenges. With a VPN connection, all computing is still done locally. Users pull data down to their laptop, but there are security risks. All data stored on the laptop is exposed to the home network, and to anyone who has access to that network.

With VDI, no data is stored locally. Everything sent across the network is a video stream. Organizations using VDI with vGPU were able keep their full team productive and secure, easily and quickly. As an example, one organization in the ship building industry needed to execute design reviews that involved very large files, 20-25GBs. Using vGPU and VMware technology, they continued reviews from home, seamlessly – impossible without the vGPU technology.

Consider it takes approximately 45 minutes to download a 20GB file through the VPN to a local laptop on a slow connection. And then, there are multiple copies of that data on laptops in homes. In the VDI with vGPU environment, users can open the same file in two minutes and the data stays in the data center. Users have the same compute resources as in the office – high-end CPUs, high-end GPUs, very fast RAM. They get a workstation-level experience over a 10MB connection at home. VDI with vGPUs is essential for speed and security for workers handling sensitive information in large files.

On the knowledge worker side, we worked with an agency who has thousands of concurrent users. In early 2019, they were unsure if knowledge workers needed GPUs. But, as we dug into how they worked, many of these users would regularly embed video into their PowerPoint presentations to communicate about their missions. We did a lot of testing and considered video playback and conferencing needs. The vGPUs made a big difference. With this environment in place, the agency easily shifted to a mass telework when the pandemic hit. They can maintain that vital level of communication across their teams and departments.

MeriTalk: What’s next for Federal telework?

Carr: Federal remote work is not going away anytime soon and will continue to expand. Applications are going to continue to get more graphically and compute intensive and more complex, the data is going to continue to get bigger, and resolutions will continue to get smaller and smaller so that images become higher quality. For example, iPhone pixel resolutions have dramatically increased over just a few years. The iPhone 5 was 1.2 megapixels and the iPhone 11 is a 12 megapixel camera and it can do video at 4K. That’s a 10x increase in the last 7 years. People will continue to use more video, whether for training, collaboration, or as a part of their daily work. Take for example, PowerPoint presentations, the application itself has changed dramatically just over the past few years. Previously presentations would consist of mostly text and maybe some simple, basic images. But now the capabilities have increased, people leverage more video, more sophisticated imagery, and add audio.

That cultural change has happened, so telework is here to stay. We are going to see significantly increased telework, even once the pandemic is finally past. The future will be a mix of home and traditional office, so the two need to seamlessly support moving between them without impediment to productivity. VDI with NVIDIA vGPU supports this so they can work from home or the office with an end user experience of equal performance and access. Employees are productive and happy at home, and we want employees to have that flexibility. Regardless of what they’re doing and where they’re working, employees will get the same performance and have access to the same applications that they did when they were working in the office. The way they can achieve this is with NVIDIA vGPU technology – this enables their data centers to power the workloads of the future.