Low-Cost Single-Node GKE Cluster (GCP)

While experimenting with Google Kubernetes Engine (GKE), minimizing cloud infrastructure costs is essential. By default, Google Cloud prompts you to create a regional Autopilot cluster with a minimum of 3 nodes.

By provisioning a Standard Zonal Cluster with a single Spot (Preemptible) VM and disabling non-essential logging and monitoring, you can run a functional GKE sandbox cluster for approximately ~$0.11/hour.

---

Provisioning via gcloud CLI

export CLUSTER_NAME="gke-sandbox"
export PROJECT_ID="my-gcp-project"
export ZONE="us-west1-a"
export MACHINE_TYPE="e2-small"
export DISK_SIZE="30"

# Create single-node Spot GKE cluster
gcloud container clusters create $CLUSTER_NAME \
  --project $PROJECT_ID \
  --zone $ZONE \
  --num-nodes 1 \
  --machine-type $MACHINE_TYPE \
  --disk-type "pd-standard" \
  --disk-size $DISK_SIZE \
  --spot \
  --enable-ip-alias \
  --release-channel "regular" \
  --addons HorizontalPodAutoscaling,HttpLoadBalancing,GcePersistentDiskCsiDriver \
  --no-enable-managed-prometheus \
  --logging=SYSTEM \
  --monitoring=SYSTEM

---

Connect to Cluster

gcloud container clusters get-credentials $CLUSTER_NAME --zone $ZONE --project $PROJECT_ID
kubectl cluster-info

---

Cost Optimization Checklist via GCP Console

• Cluster Mode: Select Standard (not Autopilot).
• Location Type: Select Zonal (single zone, avoids multi-AZ control plane replication fees).
• Node Pool:
• Machine type: e2-small (or e2-medium for heavier workloads).
• Enable Spot VMs.
• Boot disk: 30 GB Standard Persistent Disk (pd-standard).
• Features Tab: Disable Managed Prometheus, Workload Vulnerability Scanning, and verbose application logging if not required.