AKS Now Manages DRA for GPU Node Pools: How to Get the Same GPU Scheduling on EKS, GKE or Your Own Kubernetes
Blog post from Qovery
AKS has introduced managed Dynamic Resource Allocation (DRA) for NVIDIA GPU node pools as a preview, allowing it to install and maintain the GPU driver, but the feature requires Kubernetes 1.35 or later and does not yet support cluster autoscaling or node auto-provisioning. DRA is an upstream Kubernetes API that became generally available in version 1.34, replacing simple GPU counts with claims that can select devices by attributes such as model, memory, and hardware locality; the open-source NVIDIA DRA driver can provide this capability on AKS, EKS, GKE, and self-managed clusters. The article explains that cloud-specific setup varies: EKS offers driver support in Auto Mode and installation options for other node types, while GKE and self-managed clusters generally require teams to install and maintain the driver, and each platform has limitations around autoscaling or GPU sharing. For portability, it recommends dedicated DRA GPU node pools, using a consistent DeviceClass and ResourceClaimTemplates, and keeping cloud-specific configuration out of application manifests; the device plugin and DRA should not run on the same node. DRA is most useful when workloads need device-specific selection, fallback between GPU types, or per-device fault handling, while the device plugin may remain preferable for production workloads that depend on unsupported autoscaling or GPU-sharing features. Qovery currently schedules GPUs through the NVIDIA device plugin and does not configure DRA directly, though users can deploy the DRA driver and claim-based workloads as Helm services on clusters connected to Qovery.
| Trend | Post Mentions | Total Month Mentions | Posts | Companies | MoM |
|---|---|---|---|---|---|
| Kubernetes | 62 | 3,388 | 294 | 94 | -34% |
| MCP | 9 | 7,404 | 711 | 183 | -38% |
| Platform Engineering | 3 | 1,080 | 220 | 61 | -46% |
| AI Agents | 2 | 4,689 | 923 | 218 | -33% |
Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.