One pod, one node: The architecture behind serverless Kubernetes on AWS
The article explains how AWS EKS Fargate achieves serverless Kubernetes by making the pod the unit of infrastructure rather than the node. Key design choices include admission-time routing to prevent default scheduler interference, Fargate-specific scheduling for capacity provisioning, network isolation through separate namespaces, and runtime images supplying only necessary Kubernetes components. This architecture eliminates customer responsibility for creating, patching, scaling and maintaining worker nodes.
Key points
- EKS Fargate provisions dedicated compute per pod with automatic teardown
- Admission routing and Fargate-specific scheduling prevent node pool conflicts
- Network namespaces isolate platform and workload paths for enhanced security
Summary and key points written by AI from the headline and a short excerpt, so they can be incomplete or wrong. The New Stack did the reporting; read the original for the full story.
Read the full story at The New Stack
