Tutorial: Unlock the Future of Kubernetes and Accelerators With Dynamic Resource Alloc... Rey Lejano

Description

Don't miss out! Join us at our next Flagship Conference: KubeCon + CloudNativeCon events in Amsterdam, The Netherlands (23-26 March, 2026). Connect with our current graduated, incubating, and sandbox projects as the community gathers to further the education and advancement of cloud native computing. Learn more at https://kubecon.io Tutorial: Unlock the Future of Kubernetes and Accelerators With Dynamic Resource Allocation (DRA) - Rey Lejano, Red Hat At the heart of the AI revolution are GPUs and the platform that provides access to them is Kubernetes. Workloads historically access GPUs and other devices with the device plugin API but features are lacking. The new Dynamic Resource Allocation (DRA) feature helps maximize GPU utilization across workloads with additional features like the ability to control device sharing across Pods, use multiple GPU models per node, handle dynamic allocation of multi-instance GPU (MIG) and more. DRA is not limited to GPUs but any specialized hardware that a Pod may use including network attached resources such as edge devices like IP cameras. DRA is a new way to request for resources like GPUs and gives the ability to precisely control how resources are shared between Pods. This tutorial introduces DRA, reviews the “behind-the-scenes” of DRA in the Kubernetes cluster and walks through multiple ways to use DRA to request for GPU and a network attached resource.