Red Hat brings OpenShift to NVIDIA BlueField DPUs for AI boost
Red Hat has announced its support for Red Hat OpenShift on NVIDIA BlueField data processing units (DPUs), targeting enhanced efficiency, security, and performance for artificial intelligence workloads in modern data centres.
Technical preview
Red Hat OpenShift on NVIDIA BlueField DPUs is positioned to simplify the deployment of demanding AI application workloads by providing optimised security, networking, and storage features. The technical preview support is expected to be available in the coming weeks.
Organisations deploying AI applications often face the issue of resource contention between critical infrastructure services and AI workloads. This contention can result in performance limitations and increased security vulnerability. With its new support, Red Hat aims to streamline these environments by offering a single, cloud-native platform that can manage complex networking and infrastructure isolation requirements.
Key features
The integration introduces several advantages by offloading critical infrastructure services, such as networking and security processing, from CPUs to the DPUs. Red Hat OpenShift's resource management now frees up CPU capacity by shifting networking tasks to the BlueField DPU. This is intended to optimise resource usage and enable stronger AI application performance.
The solution also promises to accelerate data plane and storage traffic through the offloading of cryptographic and storage tasks. It includes support for protocols such as NVMe over Fabrics (NVMe-oF) and an accelerated Open vSwitch (OVS) data path, designed to enhance performance for data-intensive workloads.
Additional highlights are included in networking and security. The distributed routing capabilities of BlueField are expected to enhance multi-tenancy, scalability, and operational efficiency across distributed cloud clusters. Isolating infrastructure tasks at the hardware level is considered to help reduce attack surfaces and improve the overall security posture for these environments.
Ongoing collaboration
In addition to the technical preview, Red Hat noted ongoing collaboration with NVIDIA to further integrate the NVIDIA DOCA software framework and additional third-party network capabilities on OpenShift. Efforts are also underway to support NVIDIA's Spectrum-X Ethernet networking hardware, intended to support high-performance connectivity for AI workloads that span distributed cloud environments.
Ryan King, Vice President, AI and Infrastructure, Partner Ecosystem Success, Red Hat, said:
"As the adoption of generative and agentic AI grows, the demand for advanced security and performance in datacenters has never been higher, particularly with the proliferation of AI workloads. Our collaboration with NVIDIA to enable Red Hat OpenShift support for NVIDIA BlueField DPUs provides customers with a more reliable, secure and high-performance platform to address this challenge and maximise their hardware investment."
The NVIDIA BlueField-4 hardware is designed to deliver enhanced capabilities through next-generation acceleration and increased integration with DOCA, supporting the evolving cloud-native landscape and emerging AI factory models.
Justin Boitano, Vice President, Enterprise Products, NVIDIA, said:
"Data-intensive AI reasoning workloads demand a new era of secure and efficient infrastructure. The Red Hat OpenShift integration of NVIDIA BlueField builds on our longstanding work to empower organisations to achieve unprecedented scale and performance across their AI infrastructure."
Industry context
As enterprises increasingly move to deploy AI workloads at scale, the ability to support secure, high-performance, and multi-tenant environments is a growing requirement. Hardware offloading of network and security processing has become an important area of investment, as organisations look to maximise their investment in core compute resources while safeguarding sensitive data and ensuring operational efficiency.
Red Hat's latest move with NVIDIA follows a wider industry trend towards the adoption of DPUs and smart networking technologies to streamline infrastructure operations and minimise risks associated with large-scale AI deployments.
