DRA · Dynamic Resource Allocation
The Kubernetes mechanism that replaces counting opaque devices with claiming them by attribute, letting a workload request GPUs by memory, topology, or partition rather than by integer count.
Current numbers
GA in 1.34Kubernetes Dynamic Resource Allocation graduated to stable (Sept 2025)
Apache-2.0, Apr 2025NVIDIA open-sourced KAI Scheduler (gang, fair-share, DRA, topology)
1 / nodemax IMEX channels (ComputeDomains) per node in Kubernetes DRA — strands partial-node GPUs