Skip to content

Latest commit

 

History

History
75 lines (54 loc) · 2.18 KB

ram.md

File metadata and controls

75 lines (54 loc) · 2.18 KB

K8s

RAM Usage Analysis

Memory Requirements for Minimal HA Setup

This document estimates the RAM (memory) usage of Kubernetes components in a minimal High Availability (HA) setup. It includes the memory requirements for control plane nodes, worker nodes, and shared tools, and it calculates the total memory footprint for the entire cluster.


Key Kubernetes Components

Control Plane Node

  1. kube-apiserver
    • Memory usage: ~512 MB
  2. kube-controller-manager
    • Memory usage: ~200 MB
  3. kube-scheduler
    • Memory usage: ~100 MB
  4. etcd
    • Memory usage: ~256 MB (varies with cluster size)

Total Control Plane Components RAM Usage: ~1,068 MB (~1.1 GB)


Worker Node

  1. kubelet
    • Memory usage: ~512 MB
  2. kube-proxy
    • Memory usage: ~64 MB

Total Worker Node Components RAM Usage: ~576 MB (~0.6 GB)


Shared Components Across All Nodes

  1. Container Runtime (e.g., containerd or CRI-O)
    • Memory usage: ~128 MB
  2. Networking Plugin (e.g., Calico, Flannel)
    • Memory usage: ~64 MB
  3. Additional Tools (e.g., kubectl, Helm)
    • Memory usage: ~32 MB

Total Shared Components RAM Usage per Node: ~224 MB (~0.2 GB)


Total RAM Estimation

Per Node

  1. Control Plane Node:

    • Kubernetes components: ~1,068 MB
    • Shared components: ~224 MB
    • Total: ~1,292 MB (~1.3 GB)
  2. Worker Node:

    • Kubernetes components: ~576 MB
    • Shared components: ~224 MB
    • Total: ~800 MB (~0.8 GB)

HA Setup (3 Control Plane + 2 Worker Nodes)

  • Control Plane Nodes (3 × 1,292 MB): 3,876 MB (~3.9 GB)
  • Worker Nodes (2 × 800 MB): 1,600 MB (~1.6 GB)

Total RAM for HA Setup: ~5,476 MB (~5.5 GB)


Notes

  • These estimates cover the memory usage of Kubernetes components only. Additional memory will be needed for application workloads, monitoring tools, and logging systems.
  • Recommended memory for nodes is typically higher to provide headroom for operational stability:
    • Control Plane Node: 2 GB
    • Worker Node: 1 GB
  • Real-world memory usage may vary depending on cluster size, workloads, and configuration settings.