NVIDIA Brings Live GPU Migration, Ultra-High-End Workstation Performance to Virtualization

by Anne Hecht

They say good things come in threes. That makes it a banner day for the advancement of virtual desktop infrastructure, as NVIDIA has announced:

  • Availability of new virtualization software capabilities with the NVIDIA virtual GPU March 2018 Release, including improved data center management with support for live migration of GPU-accelerated virtual machines.
  • NVIDIA Quadro Virtual Data Center Workstation (Quadro vDWS) support for Tesla V100 GPUs.
  • Enhanced NVIDIA GRID Virtual PC (GRID vPC) with support for multiple 4K monitors and Linux.

The announcements came at the GPU Technology Conference, taking place through March 29 in San Jose.

Live Migration Keeps VDI Deployments Up and Running

Live migration saves valuable time and resources by allowing IT teams to focus on more strategic projects and drive business transformation.

Using Citrix XenMotion, IT teams can migrate live, NVIDIA GPU-accelerated VMs with no impact to users. And with VMware vSphere, they can suspend desktops and resume them later on compatible infrastructure, while preserving desktop and application states.

Live migration not only eases routine maintenance, it facilitates proactive maintenance — IT can resolve potential problems before service disruption occurs. Manual load balancing enables IT to optimize the utilization of resources without end user disruption.

And the best part: IT teams can perform updates and patches more frequently, at their own time. Keeping servers in healthy states is particularly difficult with today’s complex VDI environments. Without live migration, IT teams can spend hours handholding and coordinating between groups — and nights and weekends planning and performing upgrades to ensure users continually receive a great experience with minimal disruption.

Limited resources and regular quarterly and semi-annual updates recommended by Citrix and Microsoft, respectively, to keep servers secure often leave IT teams in a “keep the lights on” trap. With live migration, they can perform these updates easily, ensuring that they get the most out of their investments, without disrupting users.

Industry Support for Live Migration

In January, NVIDIA, Citrix and VMware kicked of a tech preview of live migration and early reviews are in:

“It’s a game changer. The Meltdown and Spectre BIOS came out one day, and the next day I was able to immediately make it a simple, during the workday, BIOS update. Normally this would be a 10-11 p.m. task on my own time,” says Jeremy Stroebel, IT director at Browning Day Mullins Dierdorf (BDMD)

“Now we don’t have to worry about the large Linux VMs that researchers use to run jobs overnight getting interrupted.  Students can now work on an application, go to another class, and get back to the state where they were even if we had to perform maintenance. With VMware suspend/resume for GPU-accelerated VMs, we can tighten the maintenance window for our quarterly and monthly updates or do more inside the same window,“ says Jon Kelley, associate director of enterprise innovation at University of Arkansas.

“XenServer patching was once a nightmare, taking a month or requiring several hours of downtime. With the live migration feature from Citrix and NVIDIA, I can provide a great user experience and still have the flexibility to keep my environment more secure and stable. And I can even do it within working hours without downtime,” says Tommy Stylsvig Würtz Rasmussen, IT specialist at Holstebro Kommune.

“XenMotion support on vGPU is very critical as we are a global company running almost 24 hours a day with a really short service window. Now I have the flexibility to take a host out of production without any issue. With today’s security issues like Meltdown and Spectre this is a must have to keep our environment patched and safe!” says Rody Kossen, senior system engineer at AWL-Techniek B.V.


The Most Powerful Virtual Workstation Gets More Powerful

Built to accelerate deep learning, HPC and graphics, the NVIDIA Tesla V100 is the world’s most advanced data center GPU ever. The new NVIDIA Quadro vDWS with Tesla V100 GPU support is the most powerful data center workstation. With it, users can:

  • Run real-time, interactive simulations up to 55 percent faster than previous generations such as Ansys Discovery Live.
  • Speed rendering time of photorealistic images up to 80 percent faster than with previous generations.
  • Leverage deep learning-enhanced applications for more fluid, visual interactivity throughout the design process.

Accessed from any connected device, from any location, the latest Quadro vDWS delivers even more advanced professional workstation features to designers and technical professionals, freeing creativity to happen anywhere.

“With the massive parallel computing power of NVIDIA Volta GPU architecture we’ve been able to harness deep learning to train our artificial neural network to predict travel times for transportation networks, helping our customers produce predictions 2-3 times faster than before. And with Quadro vDWS support for GV100, we have the flexibility to run VDI during the day and deep learning at night, which helps maximize use of our compute resources while keeping data secure in the data center,” says John Meza, performance engineering and virtualization team leader, ESRI

Delivering More Value to More Enterprise Users

The NVIDIA virtual GPU March 2018 release also brings enhanced capabilities with GRID vPC – including support for multiple 4K-resolution monitors and larger frame buffer support to enable increased productivity and multitasking. These features are critical to industries like healthcare and financial services, as well as for today’s knowledge workers.

With Linux application support, GRID vPC can be used by engineers who work in 2D and use electronic design automation tools, as well as software developers who code in Linux-based software development environments.

Experience NVIDIA Virtual GPU Products

See GPU-accelerated live migration at the GPU Technology Conference, where global architectural firm Browning Day Mullins Dierdorf will share how it deployed a GPU-accelerated VDI with live migration. Also on display, the NVIDIA Quadro vDWS running on Tesla V100.

See a complete listing of virtualization sessions at GTC.

Get more information about live migration for GPU-accelerated VMs.

Join our webinar, What’s New with NVIDIA Virtual GPU, to learn more.

Learn more about NVIDIA vGPU solutions by following @NVIDIAVirt.