How to Build Multi‑Cluster Continuous Delivery with ACK One and Cloud Effect

Overview

In the cloud‑native era, growing business demands, regional resource limits (e.g., GPU shortages), disaster‑recovery needs, and compliance requirements lead many users to adopt multiple Kubernetes clusters. Alibaba Cloud’s Distributed Cloud Container Platform ACK One provides a multi‑cluster application distribution feature that can dispatch resources across clusters using various strategies (static weight, dynamic scheduling, gang scheduling, etc.) while supporting intelligent resource scheduling, rescheduling, and fault migration.

Key Capabilities of ACK One Multi‑Cluster Application Distribution

Intelligent multi‑region, multi‑cluster resource scheduling : Finds resources across clusters and elastic ECS/Serverless instances, maximizing utilization and minimizing stock‑outs.

Multi‑cluster gang scheduling : Pre‑allocation or dynamic detection ensures workloads are scheduled together across clusters.

Multiple distribution strategies : Includes copy, static weight, dynamic weight, etc.

Rescheduling and fault migration : Failed workloads are reclaimed and redeployed to other suitable clusters.

Workflow

Submit application resources and a PropagationPolicy to the fleet.

The fleet’s Global Scheduler selects appropriate clusters based on the policy and distributes the resources.

If a workload fails, the Global Scheduler rescues the job and performs rescheduling or fault migration to a cluster with sufficient resources.

Building Multi‑Cluster CD with Cloud Effect

Cloud Effect is Alibaba Cloud’s one‑stop DevOps platform. By combining Cloud Effect’s CD capabilities with ACK One’s multi‑cluster distribution, you can transform an existing single‑cluster CD pipeline into a multi‑cluster CD system without modifying the original application YAML.

Steps:

Create an ACK One fleet and register multiple clusters (public‑cloud or IDC K8s clusters) to the fleet.

Add PropagationPolicy and, if needed, OverridePolicy YAML files to the Codeup repository; the original YAML remains unchanged.

Update the single‑cluster Cloud Effect pipeline to deploy the application and policies to the ACK One fleet instead of a single cluster.

Replace the single‑cluster kubeconfig with the fleet’s kubeconfig; the fleet ID is used as the cluster ID.

Run the pipeline. The application and policies are deployed to the fleet, which then distributes them to the target clusters according to the defined policies. Existing single‑cluster workloads continue running without interruption or pod restarts.

Summary

ACK One’s multi‑cluster application distribution offers a unified way to manage applications across clusters, reducing management cost and improving resource utilization. By adding only two items—distribution policies in the Git/Codeup repository and switching to the fleet kubeconfig —you can quickly convert a single‑cluster CD pipeline into a robust multi‑cluster continuous delivery solution.