Infrastructure · Software component
GPU Device Plugin
Software componentInfrastructureInfrastructurearc:AcceleratorDevicePlugin
A node-level agent that advertises GPU devices, including hardware partitions, to the container orchestrator as schedulable resources.
Responsibility. Makes GPUs visible and allocatable to container scheduling.
Also known as: NVIDIA device plugin for Kubernetes, GPU device plugin, GPU Device Plugin
Relationships
deployed on structural
sends data to dynamic
is orchestrated by control
Design guidance
- SHOULD advertise each partition profile as a distinct schedulable resource (e.g., nvidia.com/mig-1g.10gb) so workloads request a specific size.
Quantitative guidance
As stated by the sources; verify before use.
- Time-slicing example shares one GPU among 4 containers (Ref4.07).
Classification
- Patterns
- Extended resource advertisementGPU time-slicing (N containers share one GPU)Container Device Interface (CDI) injection
- Technologies
- NVIDIA device plugin for KubernetesNVIDIA Kubernetes device pluginContainer Device Interface (CDI)
- Quality attributes
- Performance efficiency (ISO/IEC 25010)
- Risks mitigated
- GPUs invisible to pods
Sources
- Ch4.5: T. Nguyen, "NVIDIA NIM and Triton Inference Server," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 4.5. ISBN: 9798244538229.
- Ch7.6: T. Nguyen, "Multi-Instance GPU," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 7.6. ISBN: 9798244538229.
- Ref4.07: NVIDIA, "About the NVIDIA GPU Operator," NVIDIA GPU Operator Documentation. Accessed: Sep. 27, 2026. [Online]. Available: https://docs.nvidia.com/datacenter/cloud-native/gpu-operator/latest/index.html