GKE Scaling Guide

Complete guide to autoscaling in Google Kubernetes Engine (GKE), covering Horizontal Pod Autoscaler (HPA), Vertical Pod Autoscaler (VPA), and Cluster Autoscaler.

Scaling Overview

Scaling Types

Autoscaler	Scales	Based On	Use Case
HPA	Pod replicas	CPU, memory, custom metrics	Variable traffic
VPA	Pod resources	Resource usage	Right-sizing
Cluster Autoscaler	Node count	Pending Pods, utilization	Capacity
NAP	Node pools	Pod requirements	Diverse workloads

---

Horizontal Pod Autoscaler (HPA)

Automatically adjusts the number of Pod replicas based on metrics.

HPA Algorithm

Formula: desiredReplicas = ceil(currentReplicas * (currentMetric / targetMetric))

Basic HPA (CPU)

apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
  name: my-app-hpa
spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: my-app
  minReplicas: 2
  maxReplicas: 10
  metrics:
  - type: Resource
    resource:
      name: cpu
      target:
        type: Utilization
        averageUtilization: 50

HPA with Multiple Metrics

apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
  name: multi-metric-hpa
spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: my-app
  minReplicas: 2
  maxReplicas: 20
  metrics:
  # Scale on CPU
  - type: Resource
    resource:
      name: cpu
      target:
        type: Utilization
        averageUtilization: 70
  # Scale on Memory
  - type: Resource
    resource:
      name: memory
      target:
        type: Utilization
        averageUtilization: 80
  # Scale on custom metric (requests per second)
  - type: Pods
    pods:
      metric:
        name: requests_per_second
      target:
        type: AverageValue
        averageValue: "1000"
  behavior:
    scaleDown:
      stabilizationWindowSeconds: 300
      policies:
      - type: Percent
        value: 10
        periodSeconds: 60
    scaleUp:
      stabilizationWindowSeconds: 0
      policies:
      - type: Percent
        value: 100
        periodSeconds: 15

HPA Metric Types

Type	Description	Example
Resource	CPU or memory utilization	CPU at 50%
Pods	Custom metric per Pod	Requests/sec
Object	Metric from another K8s object	Queue length
External	Metric from outside cluster	Pub/Sub backlog

HPA Behavior Configuration

---

Vertical Pod Autoscaler (VPA)

Automatically adjusts CPU and memory requests/limits.

VPA Configuration

apiVersion: autoscaling.k8s.io/v1
kind: VerticalPodAutoscaler
metadata:
  name: my-app-vpa
spec:
  targetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: my-app
  updatePolicy:
    updateMode: Auto  # Off, Initial, Recreate, Auto
  resourcePolicy:
    containerPolicies:
    - containerName: '*'
      minAllowed:
        cpu: 100m
        memory: 128Mi
      maxAllowed:
        cpu: 4
        memory: 8Gi
      controlledResources: ["cpu", "memory"]

VPA Update Modes

Mode	Behavior
Off	Only provides recommendations
Initial	Sets resources only at Pod creation
Recreate	Evicts and recreates Pods to apply
Auto	Same as Recreate (may change)

VPA Recommendations Only

apiVersion: autoscaling.k8s.io/v1
kind: VerticalPodAutoscaler
metadata:
  name: my-app-vpa
spec:
  targetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: my-app
  updatePolicy:
    updateMode: "Off"  # Only recommendations

Check recommendations:

kubectl describe vpa my-app-vpa

---

Cluster Autoscaler

Automatically adds or removes nodes based on Pod scheduling needs.

How Cluster Autoscaler Works

Enable Cluster Autoscaler

# Create cluster with autoscaling
gcloud container clusters create my-cluster \
    --enable-autoscaling \
    --min-nodes=1 \
    --max-nodes=10 \
    --num-nodes=3

# Enable on existing node pool
gcloud container node-pools update my-pool \
    --cluster=my-cluster \
    --enable-autoscaling \
    --min-nodes=1 \
    --max-nodes=10

Cluster Autoscaler Settings

Setting	Description
min-nodes	Minimum nodes per zone
max-nodes	Maximum nodes per zone
total-min-nodes	Total min across all zones
total-max-nodes	Total max across all zones

Autoscaling Profiles

# Set optimize-utilization profile
gcloud container clusters update my-cluster \
    --autoscaling-profile=optimize-utilization

Profile	Behavior
balanced	Conservative scale-down (default)
optimize-utilization	Aggressive scale-down, bin-packing

Scale-Down Conditions

Cluster Autoscaler removes nodes when:

• Node utilization below 50% for 10+ minutes
• All Pods can be rescheduled to other nodes
• No Pods blocking scale-down

Pods That Block Scale-Down

Allow Pod Eviction

metadata:
  annotations:
    # Allow autoscaler to evict this Pod
    cluster-autoscaler.kubernetes.io/safe-to-evict: "true"

---

Node Auto-Provisioning (NAP)

Automatically creates and deletes node pools to match workload needs.

Enable NAP

gcloud container clusters update my-cluster \
    --enable-autoprovisioning \
    --min-cpu=1 \
    --max-cpu=100 \
    --min-memory=1 \
    --max-memory=400

NAP Configuration

gcloud container clusters update my-cluster \
    --enable-autoprovisioning \
    --autoprovisioning-config-file=config.yaml

# config.yaml
resourceLimits:
  - resourceType: cpu
    minimum: 1
    maximum: 100
  - resourceType: memory
    minimum: 1
    maximum: 400
  - resourceType: nvidia-tesla-t4
    minimum: 0
    maximum: 10
autoprovisioningLocations:
  - us-central1-a
  - us-central1-b
management:
  autoUpgrade: true
  autoRepair: true

---

Scaling Best Practices

Combining Autoscalers

Combination	Recommendation
HPA + Cluster Autoscaler	Recommended for variable traffic
VPA (Off mode)	Use for resource recommendations
HPA (CPU) + VPA (memory)	Advanced, use with caution

Resource Requests Are Critical

# Always set resource requests for HPA to work
containers:
- name: app
  resources:
    requests:
      cpu: 100m
      memory: 128Mi
    limits:
      cpu: 500m
      memory: 512Mi

Scaling Configuration Checklist

Item	Description
Set resource requests	Required for HPA and CA
Configure PDB	Protect availability during scaling
Set HPA min/max	Prevent over/under scaling
Use stabilization	Prevent thrashing
Monitor metrics	Verify scaling decisions

---

Quick Reference

Create HPA (CLI)

# Create HPA targeting 50% CPU
kubectl autoscale deployment my-app \
    --cpu-percent=50 \
    --min=2 \
    --max=10

View Autoscaler Status

# HPA status
kubectl get hpa
kubectl describe hpa my-app-hpa

# VPA status
kubectl get vpa
kubectl describe vpa my-app-vpa

# Cluster Autoscaler status
kubectl get configmap cluster-autoscaler-status -n kube-system -o yaml

Troubleshooting

Issue	Check
HPA not scaling	Resource requests set? Metrics available?
VPA not updating	Update mode? Pod can be evicted?
CA not adding nodes	Quota? Max nodes? Pod schedulable?
CA not removing nodes	Pods blocking? PDB? System pods?

---

Last updated: December 2025