An AI-assisted plugin for EKS-to-GKE migrations with built-in governance
Google Cloud released an open-source AI-assisted plugin to automate and govern migrations from AWS EKS to GKE, replacing error-prone manual processes with structured, GitOps-native workflows and deterministic guardrails.
Enterprises migrating Kubernetes workloads from AWS EKS to Google Cloud’s GKE have faced high-friction challenges, including manual dissection of infrastructure-as-code, cloud-specific architectural differences, and reliance on ad-hoc AI tools that hallucinate configurations or drop critical settings. Google’s new open-source plugin, GKE agentic migration, replaces these brittle approaches with an AI-assisted pipeline protected by deterministic guardrails, addressing issues like non-existent resource properties and lost context across interdependent files.
The plugin introduces hybrid verification by using LLM workers to author Terraform and Kubernetes YAML while running deterministic transforms for exact mappings such as Workload Identity annotations and image registries, followed by strict validations like terraform validate and Kubernetes manifest contracts before presenting changes to users. It enforces GitOps-native workflows by generating target state configurations and opening Pull Requests rather than applying changes directly to live clusters, ensuring all modifications route through standard human-in-the-loop CI/CD review processes.
Designed to protect sensitive assets, the plugin separates translation logic from data transport, generating contextual runbooks that guide teams in using purpose-built tools like Google Cloud’s Database Migration Service or Storage Transfer Service for stateful data handling. It also enables protected, asynchronous handoffs between platform and application teams by persisting long-running migration state, allowing app developers to independently translate workloads within permission-isolated folders while platform engineers establish the baseline landing zone.
Packaged as an open-source agent plugin, the tool requires no custom CLI binaries or central control planes, integrating directly into existing development workflows to deliver validated pull requests and actionable runbooks into source repositories. The plugin transforms cloud migrations into predictable, AI-assisted, and reviewable GitOps workflows, with an onboarding guide available for testing against sample EKS repositories and community engagement channels for feedback and contributions.