The demand for efficient and scalable infrastructure to support artificial intelligence workloads is driving innovation in cloud orchestration. Mirantis and Netris have announced a modern integration designed to streamline the delivery and management of Kubernetes clusters specifically tailored for AI applications. This collaboration aims to address key operational bottlenecks hindering the rapid deployment of AI infrastructure, offering a more automated and standardized approach to network provisioning and cluster delivery.
The integration focuses on automating Kubernetes cluster delivery and data center networking for AI workloads, tackling challenges like the lack of standardized deployment processes and the complexities of manual network configuration. By combining Mirantis’ Kubernetes orchestration expertise with Netris’ network automation capabilities, the companies are aiming to transform GPU clusters into repeatable, multi-tenant AI cloud products with hardware-enforced isolation and automated scaling. This approach is particularly relevant for neoclouds, telecom operators, and enterprises building out their AI capabilities.
Automating AI Infrastructure with Kubernetes and Network Automation
At the heart of this partnership is the unification of Kubernetes orchestration with network automation at the hardware level. Mirantis manages the Kubernetes lifecycle, even as Netris provides the automation, abstraction, and multi-tenancy for the network infrastructure. This allows for a more integrated and efficient workflow, reducing the time and resources required to deploy and manage AI workloads. According to a press release, the integration eliminates operational bottlenecks and provides a standardized path to cluster deployment.
Netris distinguishes itself as the first commercial network orchestration platform to integrate NVIDIA BlueField DPUs into the data center network fabric. This integration extends hardware-level tenant isolation down to the server itself, improving efficiency by reducing the reliance on CPU cores for network functions. The companies state this results in higher density, better resource efficiency, and improved return on investment without compromising security or performance. A post on X (formerly Twitter) from Mirantis highlights the ability to scale multi-tenant GPU infrastructure at a lower cost with higher utilization.
Key Capabilities of the Integrated Solution
- Automated Infrastructure Delivery: The integration automates the delivery of all infrastructure components for Kubernetes clusters, including networking on NVIDIA Spectrum-X Ethernet, NVIDIA Quantum-X InfiniBand, and NVIDIA NVLink fabrics.
- AI Performance Optimization: Automation extends to data center networks for both east-west and north-south traffic, including NVIDIA Quantum-X InfiniBand and RoCE, ensuring predictable AI performance.
- Enhanced Multi-Tenancy: DPU-based tenant networking provides greater tenant density and GPU utilization, reducing operational costs per cluster.
- Hardware-Enforced Security: Multi-tenancy is enforced at the hardware level with isolation, fault tolerance, and data security at the switch and DPU levels, optimized for regulated and sovereign workloads.
This collaboration reflects Mirantis’ composable infrastructure approach, allowing operators to select validated network technologies that meet their specific performance and operational requirements. The integration isn’t simply adding networking as an afterthought; instead, the network is now an integral part of the Kubernetes cluster delivery process. As noted in a LinkedIn post from Netris, the integration delivers native hard tenant isolation.
Mirantis also recently highlighted its work with NVIDIA BlueField, delivering secure multi-tenant networking for AI. According to the Mirantis blog, they are among the first partners to deliver secure multi-tenant networking for AI with NVIDIA BlueField, turning conventional computing environments into a secure, high-performance virtual private cloud.
The integration between Mirantis and Netris represents a significant step towards simplifying the deployment and management of AI infrastructure. As AI workloads continue to grow in complexity and scale, solutions that automate and optimize the underlying infrastructure will be crucial for organizations looking to unlock the full potential of this transformative technology. The focus on hardware-level isolation and automation promises to deliver not only performance gains but also enhanced security and cost efficiency.
Looking ahead, the continued development of integrated solutions like this will be essential for accelerating the adoption of AI across various industries. The ability to seamlessly orchestrate Kubernetes and network resources will empower organizations to innovate faster and deploy AI applications with greater confidence. Share your thoughts on the future of AI infrastructure in the comments below.
