Back to Compute

Google Kubernetes Engine (GKE)

gcp

Enterprise-grade managed Kubernetes service with automated node management, security, and scaling.

COMPUTE
Google Cloud Platform

Enterprise-grade managed Kubernetes service with automated node management, security, and scaling.

Click any section above to jump directly to it

What is Google Kubernetes Engine (GKE)?

Google Kubernetes Engine (GKE) is the industry's premier managed Kubernetes platform, created by the original authors of Kubernetes at Google. GKE comes in two modes: GKE Autopilot (Google provisions and manages all nodes, security, and OS hardening while you pay purely per Pod resource request) and GKE Standard (full control over underlying node pools, machine types, and custom kernel parameters).

Decision Guide: When to Choose This Service

When to Use Google Kubernetes Engine (GKE)

Use GKE for complex microservice architectures, enterprise container workloads requiring Kubernetes APIs, service mesh (Istio), multi-container pods, stateful workloads with StatefulSets, and hybrid-cloud deployments.

When NOT to Use (Recommended Alternatives)

Do NOT use GKE if your workload is a simple stateless HTTP API or batch script that can run easily on Cloud Run with zero Kubernetes YAML overhead.

Core Architectural Pillars

Autopilot vs Standard

Autopilot manages the entire infrastructure (zero node management, pay-per-pod); Standard gives manual control over node pools and configurations.

Dataplane v2 (Cilium/eBPF)

Ultra-fast networking and native Kubernetes NetworkPolicies without iptables performance bottlenecks.

Workload Identity

Seamlessly map Kubernetes Service Accounts (KSA) directly to Google Cloud IAM Service Accounts (GSA) with zero static secret keys.

Cluster Autoscaler & Horizontal Pod Autoscaler (HPA)

Scale pods dynamically based on CPU/RAM or custom Prometheus metrics, while Cluster Autoscaler automatically adds/removes VM nodes.