Kubernetes 1.36 Pod-Level Resource Managers: Optimizing Performance and Cost

Pod-Level Resource Managers: Advanced Resource Optimization in Production

Kubernetes 1.36 introduces Pod-Level Resource Managers as an alpha feature. This system extends the kubelet’s Topology, CPU, and Memory Managers to support pod-centric resource specifications.

Why This Matters

Traditional Kubernetes resource management requires granular per-container requests and limits, which leads to over-provisioning and waste when managing complex workloads with sidecars. The ideal model allows for dynamic sharing across a pod boundary; without this, engineers face increased infrastructure costs and performance degradation due to rigid container boundaries and suboptimal NUMA node affinity.

Key Insights

• Pod-Level Resource Management (K8s 1.36): Shifts from container-centric to pod-centric allocation for CPU, memory, and hugepages.
• In-Place Vertical Scaling (Beta in 1.36): Enables resource adjustment on running pods without recreation, eliminating downtime and loss of local state.
• NUMA Topology Optimization: Integrates with the kubelet’s topology manager to ensure CPU and memory are allocated on the same NUMA node for improved cache locality.

Working Examples

Configuration for pod-level resource requests and limits in Kubernetes 1.36.

apiVersion: v1
kind: Pod
metadata:
  name: resource-intensive-app
spec:
  resources:
    requests:
      cpu: "4"
      memory: "8Gi"
      hugepages-1Gi: "2Gi"
    limits:
      cpu: "8"
      memory: "16Gi"
      hugepages-1Gi: "4Gi"
  containers:
  - name: main-app
    image: my-app:latest
    # No container-level resource specs needed
  - name: sidecar
    image: monitoring-agent:latest
    # Resources shared from pod-level allocation

Practical Applications

References:

• https://dev.to/matthiasbruns/kubernetes-136-pod-level-resource managers-advanced-resource optimization-in production-10fa