GKE shared responsibility

This page explains the shared security responsibilities for both Google and Google Cloud customers. Running a business-critical application on Google Kubernetes Engine (GKE) requires multiple parties to have different responsibilities. Although this page is not an exhaustive list, this document can help you understand your responsibilities.

This document is for Security specialists who define, govern and implement policies and procedures to protect an organization's data from unauthorized access. To learn more about common roles and example tasks that we reference in Google Cloud content, see Common GKE user roles and tasks.

Google's responsibilities

• Protecting the underlying infrastructure, including hardware, firmware, kernel, OS, storage, network, and more. This includes encrypting data at rest by default, providing additional customer-managed disk encryption, encrypting data in transit, using custom-designed hardware, laying private network cables, protecting data centers from physical access, protecting the bootloader and kernel against modification using Shielded Nodes, and following secure software development practices.
• Hardening and patching the nodes' operating system, such as Container-Optimized OS or Ubuntu. GKE promptly makes any patches to these images available. If you have auto-upgrade enabled, or are using a release channel, these updates are automatically deployed. This is the OS layer underneath your container—it's not the same as the operating system running in your containers.
• Building and operating threat detection for container-specific threats into the kernel with Container Threat Detection (priced separately with Security Command Center).
• Hardening and patching Kubernetes node components. All GKE managed components are upgraded automatically when you upgrade GKE node versions. This includes:
  • vTPM-backed trusted bootstrap mechanism for issuing kubelet TLS certificates and auto-rotation of the certificates
  • Hardened kubelet configuration following CIS benchmarks
  • GKE metadata server for Workload identity
  • GKE's native Container Network Interface plugin and Calico for NetworkPolicy
  • GKE Kubernetes storage integrations such as the CSI driver
  • GKE logging and monitoring agents
• Hardening and patching the control plane. The control plane includes the control plane VM, API server, scheduler, controller manager, cluster CA, TLS certificate issuance and rotation, root-of-trust key material, IAM authenticator and authorizer, audit logging configuration, etcd, and various other controllers. All of your control plane components run on Google-operated Compute Engine instances. These instances are single tenant, meaning each instance runs the control plane and its components for only one customer.
• Provide Google Cloud integrations for Connect, Identity and Access Management, Cloud Audit Logs, Google Cloud Observability, Cloud Key Management Service, Security Command Center, and others.
• Restrict and log Google administrative access to customer clusters for contractual support purposes with Access Transparency.

Customer's responsibilities

• Maintain your workloads, including your application code, build files, container images, data, Role-based access control (RBAC)/IAM policy, and containers and pods that you are running.
• Rotate your clusters credentials.
• Keep Standard node pools enrolled in automatic upgrades.
• In the following situations, manually upgrade your clusters and node pools to remediate vulnerabilities within your organization's patching timelines:
  • Auto-upgrades are postponed because of factors like maintenance policies.
  • You need to apply a patch before it becomes available in your selected release channel. For more information, see Run patch versions from a newer channel.
• Monitor the cluster and applications and respond to any alerts and incidents using technologies such as the security posture dashboard and Google Cloud Observability.
• Provide Google with environmental details when requested for troubleshooting purposes.
• Ensure Logging and Monitoring are enabled on clusters. Without logs, support is available on a best-effort basis.

What's next

• Read the GKE Security overview.