At VMworld: NVIDIA vGPU with vMotion Delivers the Agile Data Center

VMware and NVIDIA open tech preview of vMotion for GPU Live Migration.
by Anne Hecht

VMware and NVIDIA announced at VMworld 2018 today a technology preview of vSphere vMotion for NVIDIA GPUs with VMware vSphere 6.7 update 1, offering early access to select participants.

vMotion’s live migration capability has been the most requested feature among customers since NVIDIA virtual GPU (vGPU) was first released in 2013. General availability is expected later this year.

Keeping VDI Users Productive, and Servers Healthy

IT admins with access to the tech preview of NVIDIA vGPU with VMware vSphere 6.7u1 can use vMotion to live migrate NVIDIA vGPU-powered virtual desktops without end-user interruption and no data loss.

This allows IT to perform critical services like workload leveling, infrastructure resilience and server software upgrades without any VM downtime, when it’s convenient for them. With live migration, IT can truly deliver high availability and a quality user experience.

NVIDIA-Powered Data Center Never Sleeps

With NVIDIA GPUs in the data center, businesses can run multiple types of workloads — whether it’s Monte Carlo simulations for assessing financial investments or 3D graphics and data-intensive workloads for oil and gas exploration. This enables data center admins to maximize data center utilization, improving productivity and reducing costs.

When a VM goes idle, IT can consolidate the remaining VMs by live migrating them to a different host. And they can repurpose the original host to run compute workloads, like HPC and deep learning, at night.

This is possible because NVIDIA’s virtualization software — NVIDIA Quadro Virtual Data Center Workstation (Quadro vDWS) and NVIDIA GRID — run on the same Tesla Volta and Pascal GPUs as deep learning, inferencing, training and HPC workloads. When graphics resources are needed for VDI the next day, IT admins can simply repurpose the NVIDIA GPUs to virtual GPUs to support VDI again.

“In the past, GPU virtualization was limited to developers on remote sites and automated testing, where a little planned downtime is acceptable,” said Ouissame Bekada, system, virtualization and storage engineer at Dassault Systèmes. “With vMotion technology support on NVIDIA virtual GPU, we can now expand virtualization to our production environments, were failure and downtime is not an option. Patching, load balancing, and upgrades of hardware and software have become a breeze, and end-users don’t notice a thing.”

“GPUs have revolutionized the way we work by enabling users to run virtualized AI-accelerated and deep learning-enhanced applications to modern productivity and collaboration applications,” said Courtney Burry, vice president of product marketing for End-User Computing at VMware. “This makes VMware vSphere vMotion for NVIDIA vGPUs more relevant than ever, allowing IT to maximize end-user productivity and optimize infrastructure investment. This can help solve one of the most difficult challenges in the virtualization space.”

NVIDIA vGPU Innovations Showcased at VMworld

Visit NVIDIA at booth 1362 and experience virtualization for every use case — from virtualized deep learning-enhanced applications and real-time simulation to modern productivity applications. Demos include:

  • vMotion of GPU-accelerated VMs.
  • Real-time interactive simulation on Ansys Discovery Live running on Quadro vDWS and Tesla GPUs to accelerate design cycles. Use these same GPUs at night to run large HPC workloads.
  • AI-enhanced applications benefit professional applications and workflows (includes large-scale object detection).
  • The digital workplace redefined with higher density in VDI environments.
  • Advanced, AI-accelerated real-time ray-traced global illumination, shadows, ambient occlusion and reflections.

Come meet our team of product experts at the vGPU Expert Bar at VMworld. The daily schedule covers a variety of topics so you can get your toughest VDI questions answered.

Learn more about NVIDIA vGPU solutions by following @NVIDIAVirt.