Innovations like artificial intelligence (AI), machine learning (ML) and other emerging workloads present a vision of IT’s future, one where intelligent solutions can more effectively analyze and address evolving business needs. But this vision can be limited by current IT infrastructure, which can often require significant investments in order to enable new workloads.
One answer to this challenge is through workload acceleration, which uses specialized computational resources, like graphic processing units (GPUs) to tackle intense computing tasks. Established in scientific and research computing, GPUs such as those offered by NVIDIA are now catching the attention of enterprise IT as a technology that can accelerate compute-intensive operations found in data science and AI, extending their reach to a broader range of end users.
We understand this need to deliver emerging workloads on standardized, accessible infrastructure, which is why Red Hat has been diligently working on enabling NVIDIA® GPU-based acceleration in our enterprise platforms, including Red Hat Enterprise Linux and Red Hat OpenShift. Additionally, our expanded collaboration with NVIDIA is intended to help drive technologies and solutions for AI, enterprise high performance computing and data sciences across different industries, and we continue to further this work across Red Hat’s portfolio of enterprise-ready open source technologies.
Today, we are announcing the availability of Red Hat Enterprise Linux and Red Hat OpenShift on NVIDIA DGX-2 systems, which are designed to deliver powerful solutions for complex AI challenges. This will help to bring accelerated data science, ML and deep learning (DL) workflows to the Linux-based heart of enterprise datacenters and enable customers to more effectively design and execute demanding AI applications in their existing environments. Red Hat OpenShift, the industry’s most comprehensive enterprise Kubernetes platform, helps to facilitate deployments across the hybrid cloud, from on-premises servers to public cloud environments, enabling customers to migrate applications more seamlessly across IT footprints.
Red Hat and NVIDIA are seeking new ways to enable the latest technological innovations for our ecosystem. The compact and energy-efficient NVIDIA T4 GPU can accelerate a diverse range of workloads, from deep learning training to data analytics, on mainstream server architectures. Red Hat is working with our hardware OEMs to certify server configurations that include NVIDIA T4 GPUs on Red Hat Enterprise Linux while NVIDIA provides driver support to help bring this technology to data centers around the globe, using familiar infrastructure.
"We’re expanding the portfolio of our accelerated computing solutions running on Red Hat Enterprise Linux to include NVIDIA DGX-2 systems and the new T4 GPUs," said Charlie Boyle, senior director, DGX Systems, NVIDIA."With NVIDIA’s software stack now available on Red Hat Enterprise Linux, enterprise IT organizations can better meet customer demand for end-to-end support of AI solutions in the hybrid cloud."
The NVIDIA NGC container registry provides GPU-optimized software tools for AI and HPC and Red Hat Enterprise Linux is officially supported through the NGC-Ready validation program. OpenShift provides a platform based on Red Hat technologies, to orchestrate these containers at scale and enable mutual customers to take advantage of NVIDIA GPUs in their infrastructure.
Specifically, Red Hat OpenShift provides a host of features aimed at enabling Kubernetes clusters to more effectively run accelerated AI workloads in containers. Red Hat and NVIDIA have been collaborating in the open source community to deliver features such as:
Device plugins that provide access to GPUs and other specialized hardware to applications running in containers
Huge Pages support to enable containers with large memory requirements to run more efficiently
Our collaboration is also evident in our work to improve the end user experience around the packaging, installation and maintenance of NVIDIA’s GPU drivers on Linux systems. This work helps to provide stability, reliability and an overall improved experience for users interested in deploying NVIDIA’s GPUs and CUDA software on top of Red Hat platforms in their enterprise environments.
The effort of bringing the power of CUDA development platform to our users, that began with CUDA 9.0 release, continues with the latest version - 10.1. Red Hat has collaborated with NVIDIA to bring the power of CUDA development platform to our users, helping to drive the creation of innovative applications for AI and high-performance computing use cases across Red Hat’s technology portfolio.
On the virtualization front, Red Hat and NVIDIA have been working together in the upstream community to enable support for NVIDIA virtual GPU (vGPU) software in Red Hat Virtualization, a Kernel-based virtual machine (KVM)-powered virtualization platform. We’re pleased to say that Red Hat Virtualization is the first hypervisor in the industry to support multiple vGPUs from NVIDIA in a single virtual machine. By collaborating with NVIDIA, Red Hat makes this technology available through its open, enterprise-ready virtualization platform, making it easier to drive performance and innovations across a range of industries.
Red Hat has also introduced support for vGPUs in Red Hat Hyperconverged Infrastructure for Virtualization to make it easier for organizations to increase visual clarity, improve performance and decrease lag time when delivering virtualized graphics to users at remote sites. This helps to make the remote office IT experience nearly indistinguishable from a physical desktop or workstation.
From a storage perspective, AI and ML workloads rely on data to create and train models and for post-processing. Red Hat Ceph Storage enables organizations to increase the performance of multiple workloads while reducing data redundancy and synchronization challenges. Red Hat OpenShift Container Storage, tightly integrated with our enterprise Kubernetes platform, provides persistent volumes, application portability and more seamless developer experience by serving as an abstraction layer across deployments.
We see IT’s future as one fueled by open source technologies, and we’re pleased to continue our collaboration with NVIDIA to achieve this future, from virtualized deployments in corporate datacenters to massive-scale services deployed on public clouds.
Red Hat at GTC 2019
To learn more about how Red Hat and NVIDIA align on open source solutions to fuel emerging workloads, visit Red Hat (booth #716) at the GPU Technology Conference (GTC) from March 18-21, 2019. Our team of experts will be on hand to answer questions and provide additional information about Red Hat’s product portfolio.
At the event, Red Hat and NVIDIA will also present the following sessions:
S9299 - Deploying NVIDIA vGPU with Red Hat Virtualization (RHV) - Monday, March 18, 9 a.m. to 9:50 a.m. PT in Hilton Hotel Almaden, 1 Room (Street Level)
S9292 - Red Hat and the NVIDIA DGX: Tried, Tested, Trusted - Monday, March 18, 10 a.m. - 10:50 a.m. PT in San Jose Convention Center, Room 212B (Concourse Level)
About the author
Yan Fisher is a Global evangelist at Red Hat where he extends his expertise in enterprise computing to emerging areas that Red Hat is exploring.
Fisher has a deep background in systems design and architecture. He has spent the past 20 years of his career working in the computer and telecommunication industries where he tackled as diverse areas as sales and operations to systems performance and benchmarking.