Resource Requests and Limits

Overview

Resource management ensures fair scheduling and prevents resource starvation.

Field	Purpose
requests	Minimum guaranteed resources (used for scheduling)
limits	Maximum allowed resources (enforced at runtime)

---

Resource Types

CPU:

• Measured in millicores (m) or cores
• 1 core = 1000m
• Example: 100m = 0.1 CPU

Memory:

• Measured in bytes (Ki, Mi, Gi)
• Example: 256Mi = 256 mebibytes

---

Basic Example

apiVersion: v1
kind: Pod
metadata:
  name: resource-demo
spec:
  containers:
  - name: app
    image: nginx
    resources:
      requests:
        memory: "64Mi"
        cpu: "100m"
      limits:
        memory: "128Mi"
        cpu: "200m"

---

How Scheduling Works

1. Scheduler looks at requests to find a node with enough capacity
2. Node's allocatable resources must >= sum of all pod requests
3. Limits are enforced by kubelet (cgroups) at runtime

---

Behavior When Limits Exceeded

Resource	Behavior
CPU	Throttled (process slowed down)
Memory	OOMKilled (pod terminated)

---

Best Practices

# Good: requests < limits (burstable)
resources:
  requests:
    cpu: "100m"
    memory: "128Mi"
  limits:
    cpu: "500m"
    memory: "256Mi"

# Guaranteed QoS: requests = limits
resources:
  requests:
    cpu: "100m"
    memory: "128Mi"
  limits:
    cpu: "100m"
    memory: "128Mi"

---

LimitRange

Set default/min/max for a namespace:

apiVersion: v1
kind: LimitRange
metadata:
  name: mem-limit-range
  namespace: default
spec:
  limits:
  - default:          # default limit
      memory: "512Mi"
      cpu: "500m"
    defaultRequest:   # default request
      memory: "256Mi"
      cpu: "100m"
    max:              # max allowed
      memory: "1Gi"
      cpu: "1"
    min:              # min allowed
      memory: "64Mi"
      cpu: "50m"
    type: Container

---

ResourceQuota

Limit total resources in namespace:

apiVersion: v1
kind: ResourceQuota
metadata:
  name: compute-quota
  namespace: dev
spec:
  hard:
    requests.cpu: "4"
    requests.memory: "8Gi"
    limits.cpu: "8"
    limits.memory: "16Gi"
    pods: "10"

---

Useful Commands

# Check node capacity and allocatable
kubectl describe node <node-name> | grep -A5 "Allocatable"

# Check resource usage
kubectl top pods
kubectl top nodes

# Check pod resources
kubectl describe pod <pod-name> | grep -A10 "Limits"