This comprehensive guide explains how to transform Kubernetes from a single‑cluster setup into a production‑grade, multi‑cluster platform that can handle tens of thousands of pods and high‑concurrency workloads by applying architectural, operational, and governance best practices across eight layers of the stack.
From its 2015 debut as a stateless microservice orchestrator, Kubernetes now powers large‑scale data pipelines, distributed training, high‑throughput inference, and autonomous agents, unifying these workloads on a single platform while addressing resource coordination, multi‑cluster scheduling, and GPU economics.
This article explains how GlobalElasticQuotaTree extends Kubernetes native elastic quota to multi‑cluster environments, providing hierarchical quota structures, Min/Max borrowing, cluster‑level control, and workload‑type support to improve resource utilization for AI platforms.
This article explains why AI inference services require multi‑cluster gray‑release, outlines the risks of traditional updates, and details how ACK One Fleet combined with Kruise Rollout provides a controlled, observable, and rollback‑capable solution for deploying large AI models across hybrid cloud clusters.
In the era of large AI models, ACK One’s multi‑cluster fleet provides inventory‑aware elastic scheduling, cluster‑level priority dispatch, and hybrid‑cloud strategies to maximize GPU utilization, ensure business continuity, and reduce costs across regions and on‑premise data centers.
ACK One Fleet, Alibaba Cloud's enterprise multi‑cluster solution, provides inventory‑aware elastic GPU scheduling, cross‑region resource sharing, multi‑cluster HPA and model distribution, allowing AI inference and training workloads to scale efficiently, reduce costs, and maximize GPU utilization.
This article summarizes Alibaba Cloud’s key technical contributions at KubeCon China 2025, covering AI‑focused Kubernetes optimizations, Argo Workflows enhancements, storage strategies for large models, Fluid’s data orchestration, multi‑tenant security, and the RoleBasedGroup framework for PD‑separated AI inference.
In the cloud‑native era, Kubernetes is transforming from a container orchestrator into a full‑stack operating system, with 2025 bringing AI‑native scheduling, integrated security, multi‑cluster and edge convergence, platform‑engineered developer experiences, and tight WebAssembly integration.
This article examines how to centrally manage GPU resources across heterogeneous Kubernetes clusters using namespace‑based RBAC isolation, virtual control‑plane solutions like vcluster, and multi‑cluster tools such as Karmada, comparing their architectures, use cases, advantages, and limitations to guide enterprise‑level deployment decisions.
Learn how to define and manage multiple Kubernetes clusters using contexts, extract certificates, configure contexts for production and test environments, and boost switching efficiency with the kubectx tool, complete with step‑by‑step commands and practical tips.
This guide explains how Alibaba Cloud's ACK One multi‑cluster application distribution works, outlines its key capabilities such as intelligent resource scheduling and gang scheduling, and shows step‑by‑step how to convert a single‑cluster CD pipeline into a multi‑cluster continuous delivery system using Cloud Effect.
This guide explains how to leverage Alibaba Cloud's ACK One multi‑cluster application distribution together with the Cloud Effect DevOps platform to convert an existing single‑cluster continuous delivery pipeline into a resilient, multi‑region, multi‑cluster CD solution without modifying original YAML resources.
The article analyzes the pain points of managing several Elasticsearch clusters separately, compares native Kibana, custom scripts, and commercial tools, and then walks through a practical implementation using the lightweight INFINI Console to achieve unified, version‑agnostic monitoring and alerting.
Over two years, our Application Center Platform evolved through four stages—initial KubeVela‑based delivery, added bottom‑up synchronization for legacy services, a standardized OAM‑aligned application model, and multi‑active governance with delivery rules—boosting hybrid‑cloud delivery efficiency by over 60% and enabling automated clustering, resource detection, and compliance.
This article explains the theory and practical usage of k8gb, a Kubernetes Global Balancer that provides DNS‑based load balancing, fault‑tolerant traffic routing, and seamless failover across multiple clusters to improve resilience, latency, and compliance in cloud‑native environments.
Wondershare Technology adopted Alibaba Cloud's ACK One GitOps platform to automate and unify the deployment of dozens of Kubernetes clusters across multiple regions, addressing manual deployment inefficiencies, traceability, rollback challenges, and multi‑tenant permission management while achieving a 50% increase in release efficiency.
In this interview, CNCF ecosystem lead Taylor Dolezal shares ten practical recommendations for Kubernetes architects, covering security, resource optimization, observability, high‑availability, multi‑cluster networking, stateful workloads, CI/CD, lifecycle upgrades, documentation, and a learning roadmap for the next one to two years.
This guide explains how ACK One enables fast, Git‑free distribution of Kubernetes application resources from a single fleet cluster to multiple target clusters, covering the required CRDs, policy definitions, step‑by‑step YAML examples, and migration from single‑cluster to high‑availability multi‑cluster deployments.
At KubeCon + CloudNativeCon 2024 in Hong Kong, JD Cloud presented its cross‑cluster, large‑scale application management practice, detailing a federated Serverless model that oversees over 10,000 nodes, improves resource utilization, simplifies multi‑cluster scheduling, and offers efficient elastic scaling solutions.
Version 1.0.0 of KubeAdmiral, ByteDance’s open‑source multi‑cluster orchestration engine, introduces native Kubernetes API compatibility, advanced scheduling policies, fault‑tolerant migration, global status aggregation, and extensive hybrid‑cloud support, enabling seamless management of over 210 k machines across public and private clouds.
This article explains why organizations adopt multi‑cluster Kubernetes for high availability, hybrid‑cloud scaling, and fault isolation, outlines the preparatory steps, compares Federation v1 and v2, introduces Karmada as a CNCF project, and shares practical non‑federated deployment, monitoring, traffic management, and migration techniques with code examples.
With Kubernetes clusters expanding across multiple environments, choosing the right cross‑cluster networking method is crucial; this article compares six approaches—underlay networks, overlay CNI meshes, Submariner, Skupper, KubeSlice, and others—detailing their core principles, advantages, and limitations to help you select the best fit.
This article explains how CloudLakehouse’s Multi‑Cluster elastic architecture enables high‑concurrency, low‑latency real‑time analytics on data lakes by addressing storage‑compute separation, dynamic caching, and automated scaling, providing a cost‑effective solution for customer‑facing data products.
This article details DeWu's cloud‑native transformation since August 2021, covering multi‑cluster federation, application profiling, custom scheduling plugins, resource pre‑reservation, co‑location of online and offline workloads, cost‑saving hardware choices, multi‑cloud strategy, and the development of the KubeAI platform for AI scenarios.
This guide explains why multi‑cluster deployments are needed for high‑availability, describes Istio's flat and non‑flat network models with single or multiple control planes, and provides step‑by‑step scripts to create Kind clusters, install MetalLB, configure root CAs, deploy Istio, set up gateways, and verify regional load balancing and failover.
This guide explains how to use Alibaba Cloud's ACK One platform to connect on‑premises and public‑cloud Kubernetes clusters, configure network interconnectivity, create multi‑cluster fleets, optionally deploy applications via GitOps, and manage traffic with a multi‑cluster gateway for seamless same‑city disaster recovery.
Since August 2021, DeWu App has built a cloud‑native, multi‑cluster Kubernetes platform that uses an OAM‑style CloneSet model, Helm‑generated resources, Karmada‑based federation, custom scheduler plugins for reservation and node‑balance, offline mixing for Flink, a unified KubeAutoScaler, and a self‑built KubeAI stack, achieving significant cost cuts and improved stability while planning further middleware containerization and multi‑cloud expansion.
This article examines the motivations, benefits, and existing solutions for Kubernetes multi‑cluster management, then details Vivo's non‑federated and federated approaches, application‑centric continuous delivery, elastic scaling, unified scheduling, gray‑release strategies, and summarizes the current state and challenges.
The article explains why enterprises adopt multi‑cluster Kubernetes architectures, reviews community solutions such as Karmada, Clusternet and OCM, and details vivo’s hybrid strategy that combines a unified UI for independent clusters with Karmada‑based federation for resource distribution, elastic scaling, cross‑cluster scheduling, and gray‑release migration.
This article details how ZEEK automotive tackled rapid growth challenges by redesigning its cloud‑native infrastructure, adopting Alibaba Cloud ACK FinOps and ACK One for multi‑cluster management, and implementing cost‑visibility, intelligent allocation, and configuration checks that yielded significant resource savings and operational stability.
Since Kubernetes became the de‑facto standard, ByteDance faced scaling limits with single‑cluster setups, prompting the adoption of KubeFed V2 and later the development of KubeAdmiral, a next‑generation multi‑cluster federation system that enhances scheduling, resource efficiency, native API support, and dynamic scaling across clouds.
This article examines the origins and core challenges of Kubernetes multi‑cluster management, compares four open‑source solutions—KubeFed, Karmada, Open Cluster Management (OCM) and Gardener—by detailing their architectures, application models, scheduling and service discovery mechanisms, and discusses emerging trends such as multi‑tenant isolation, virtual clusters, and stateful‑application migration strategies.
This article explains the motivations behind Kubernetes multi‑cluster deployments, outlines common use cases such as isolation and high‑availability, and analyzes core management elements including deployment models, control‑plane architectures, network connectivity, service discovery, cross‑cluster scheduling, application model extensions, and treating clusters as resources.
This article provides a detailed overview of CNStack 2.0, explaining its cloud‑service and cloud‑component model, the cn‑app‑operator lifecycle controller, Sealer‑based build/share/run workflow, and the ability‑center white‑screen management that together simplify multi‑cluster cloud‑native application delivery.
This guide explains how to enable cross‑cluster pod communication in Kubernetes using Kube‑OVN's OVN‑IC feature, covering prerequisites, single‑node and high‑availability database deployment, automatic and manual route configuration, and cleanup procedures with concrete Docker/Containerd commands and ConfigMap examples.
Drawing from Yang Xiaofei’s SACC2022 talk, this article details Zhongtong Cloud’s two‑year journey from initial containerization to a multi‑cluster architecture, covering challenges, custom scheduler extensions, fixed‑IP handling, container crash‑site preservation, node rebalancing, application migration, cross‑cluster load balancing, and future plans for unified gateways.
This article explores KubeVela’s full‑stack declarative observability framework, detailing cloud‑native monitoring challenges, the Prism Aggregated API approach, multi‑cluster configurations, and out‑of‑the‑box addons that let developers and platform engineers seamlessly integrate, customize, and scale metrics, logs, and dashboards across heterogeneous environments.
This guide explains the purpose of Kubernetes Federation, walks through installing Helm, the kubefed controller plane, and kubefedctl, then details how to join clusters, enable resource federation, deploy sample workloads, and provides a handy command reference for multi‑cluster operations.
CNStack 2.0 is a cloud‑native PaaS platform built on Kubernetes that unifies resource and workload management, offering agile, open, and secure multi‑cluster capabilities through modular cloud services, a unified API gateway, and integration with open‑source projects such as Sealer, Emissary‑Ingress, cert‑manager, Velero, and OCM.
ACK One is a distributed cloud‑native container platform that unifies management of Kubernetes clusters across hybrid‑cloud, edge, and on‑prem environments, offering GitOps‑based multi‑cluster application distribution with ArgoCD integration and a centralized alert‑management system.
The Karmada community's Cloud Native Days China 2022 session in Nanjing gathered over 30 enterprises and developers to share multi‑cloud, multi‑cluster production practices, large‑scale testing results, and real‑world implementations from Huawei Cloud, vivo, Hurricane Engine, China Mobile, DaoCloud, and Zhejiang University, highlighting Karmada's scalability and ecosystem growth.
This tutorial walks you through installing Devtron on Kubernetes, creating additional Kind clusters, adding them as managed clusters, configuring environments, adding custom Helm repositories, and deploying a sample Spring Boot application across multiple clusters using Devtron's UI and chart groups.
Vivo’s engineering team shares their practical experience creating a Karmada‑Operator using the Operator SDK and Ansible, detailing background, deployment challenges, design choices, API and architecture, etcd management, member cluster handling, CI pipeline, and performance testing to enable robust multi‑cloud Kubernetes orchestration.
vivo created an Ansible‑based Karmada‑Operator that declaratively manages multi‑cluster deployments, etcd backup/restore, and control‑plane upgrades via custom CRDs and CI pipelines, addressing the limitations of existing tools and providing extensible, reliable, self‑healing orchestration for large‑scale Kubernetes environments.
Over the past fifteen years, instant‑messaging (IM) systems have progressed through three stages—from early high‑concurrency software, through mobile‑internet and cloud‑based solutions, to modern cloud‑native architectures—highlighting shifting development challenges, the emergence of SDK services, and the technical considerations of multi‑cluster, high‑scale deployments.
The article explains how Intuit manages hundreds of Kubernetes clusters using multi‑control‑plane architectures and Admiral to automate Istio service‑mesh configuration, illustrating the challenges of large‑scale microservice deployment and sharing a domestic bank’s experience with similar cloud‑native adoption.
This article explains why Kubernetes federation is needed for managing multiple clusters, compares the deprecated Federation v1 with the improved v2 architecture, and introduces Karmada as a modern multi‑cloud orchestration solution, complete with configuration examples, scheduling strategies, and CRD definitions.
KubeVela 1.4 introduces built‑in multi‑cluster authentication and authorization, a resource‑topology visualizer, the lightweight VelaD runtime, dozens of new plugins, and a host of workflow enhancements, while outlining the project’s roadmap for observability, workflow and application management.
Vivo describes how it built a Kubernetes‑Operator and CI pipeline to automate large‑scale, multi‑data‑center cluster deployment, modular management, and lifecycle operations using Ansible, kubeadm, and kubevirt, improving reliability, maintainability, and scalability of its Kubernetes fleets.
Version 0.7 of OpenClusterManagement introduces the DefaultClusterSet for automatic cluster grouping, upgrades the Placement API to v1beta1 with Taint/Toleration‑based scheduling, lowers the Hub cluster version requirement to 1.12, adds a Hosted deployment mode, and integrates enhanced multi‑cluster features with KubeVela 1.3 and Alibaba Cloud ACK Agile.
This guide walks through creating a two‑region, three‑center disaster‑recovery architecture using ACK One’s multi‑cluster application distribution and GTM traffic management, covering cluster setup, differential configuration, workflow control, traffic policies, and verification steps with concrete kubectl commands and YAML manifests.
This article describes Yiche's adoption of Kubernetes, detailing its master‑node architecture, multi‑cluster strategies, custom container management platform, CI/CD pipeline, monitoring with Prometheus, logging, auditing, health checks, and future plans to streamline traffic and control planes.
This guide explains how to run and switch between several Kubernetes clusters on a single server by configuring separate kubeconfig entries, adding clusters, users, and contexts, and using kubectl commands to manage each cluster efficiently.
This article explains why Kubernetes single‑cluster scalability is limited to about 5,000 nodes, introduces the concept of multi‑cluster federation, compares the legacy kubefed project with the actively maintained Karmada solution, and shows how policies and replica‑scheduling enable flexible cross‑AZ deployments and failover.
eBay’s data infrastructure engineers describe how they design and implement cloud‑native, multi‑cluster high‑availability architectures for stateful data components—covering background challenges, federated Kubernetes management, state handling, fault‑tolerance, backup, and chaos testing—to ensure reliable, scalable data services across global data centers.
This article details the architectural background, challenges, and practical solutions implemented in SOFAStack CAFE for cloud‑native multi‑cluster deployment, including a custom multi‑topology CRD, an independent federation API server, enhanced KubeFed controller features, and network proxy integration to support hybrid‑cloud scenarios.
The article explains how Intuit manages hundreds of Kubernetes clusters across multiple business units using Istio‑based service mesh and the Admiral multi‑control‑plane solution to achieve isolation, compliance, automated configuration, and fault‑tolerant cross‑cluster communication.
This article explains how Open Cluster Management's Placement API selects target clusters for workloads in multi‑cluster Kubernetes environments, covering ManagedClusterSet concepts, Placement spec fields, predicate filtering, prioritizer policies, dynamic scoring, and practical YAML examples that demonstrate weight adjustments and best‑practice scenarios.
Facing large‑scale production demands, Ctrip’s Cloud Container team designed a multi‑data‑center Service Mesh architecture that isolates failures, employs dual‑active IDC, multi‑cluster deployments, and optimized xDS push mechanisms, ensuring high availability, rapid fault recovery, and efficient rollout across hundreds of services.
This guide explains how to simulate and maintain two separate Kubernetes clusters on a single node by configuring kubeconfig files, adding clusters, users, and contexts, and switching between them to improve DevOps workflow efficiency.
This article explores emerging cloud‑native technologies such as MicroVMs, high‑performance WebAssembly, lightweight Kubernetes distributions, edge‑computing platforms, multi‑cluster management, and cloud‑native backup solutions, highlighting their advantages and the role they will play after Kubernetes.
This article explores how cloud‑native principles are reshaping DevOps collaboration, examines emerging trends in next‑generation DevOps platforms, and details how KubeVela—leveraging OAM, multi‑cluster delivery, workflow definitions, and GitOps—provides a unified, application‑centric solution for modern enterprise environments.
This article explains how a container‑driven PaaS platform built on Docker and Kubernetes can centralize resource management, enable dynamic allocation, monitoring, and high‑availability across multiple data centers, while supporting microservice deployment and continuous business operations.
This article explains why modern enterprises need multiple Kubernetes clusters—covering single‑cluster limits, hybrid‑cloud requirements, and fault‑tolerance—then compares two architectural models and reviews both Kubernetes‑centric federation and network‑centric service‑mesh solutions with practical implementation guidance.
Enterprises adopt hybrid multi‑cloud architectures driven by high‑profile security incidents and regulatory demands, leveraging Kubernetes multi‑cluster capabilities such as disaster recovery, latency reduction, isolation, fault containment, and vendor‑lock‑in avoidance, with solutions like Federation v1/v2 and KubeSphere illustrated through real‑world case studies.
Clusternet is an open‑source cloud‑native Kubernetes multi‑cluster management project that enables seamless registration, synchronization, and deployment of applications across public, private, hybrid, or edge clusters using lightweight agents and hubs, supporting various sync modes, RBAC, and kubectl plugins.
The article outlines the latest OCM 0.4.0 release—including addon registration, a new Placement API with YAML examples, the alpha‑stage clusteradm CLI—while previewing the 0.5.0 roadmap, integrations with KubeVela, ArgoCD, Clusternet, and real‑world deployments on Alibaba Cloud.
This article explains why modern enterprises adopt multiple Kubernetes clusters, covering single‑cluster capacity limits, hybrid‑cloud requirements, fault‑tolerance concerns, the benefits of multi‑cluster setups, architectural models, and community‑driven implementation patterns.
Open Cluster Management (OCM) is an open‑source, hub‑agent solution that simplifies lifecycle, configuration, and policy management for multiple Kubernetes clusters across hybrid and multi‑cloud environments, detailing its architecture, core primitives, advantages, and real‑world deployments by Ant Group and Alibaba Cloud.
The article examines Kubernetes' widespread adoption, outlines its scalability and multi‑cluster management constraints, discusses practical application scenarios such as deployment models, batch scheduling, and hard multi‑tenancy, and highlights the gaps that still limit its use in large‑scale production environments.
The article outlines Kube‑OVN’s evolution from its 2019 debut to becoming the CNCF‑hosted open‑source CNI with extensive enterprise adoption, details recent feature releases such as multi‑VLAN, BGP, Smart‑NIC, DPDK, dual‑stack and multi‑cluster support, and presents a 2021‑2022 roadmap focusing on performance, security, monitoring and broader cloud‑native use cases.
This guide introduces kubecm, a Go‑based CLI tool for simplifying multi‑cluster kubeconfig management, covering its background, cross‑platform installation, core commands, command‑line auto‑completion, and practical usage screenshots to help Kubernetes users work more efficiently.
This article details how the QingJiao Classroom platform evolved from a single‑cluster setup to a multi‑cluster architecture using KubeSphere on QingCloud (QKE), covering selection criteria, deployment strategies, automation, monitoring, authentication, elastic scaling with Keda, and future plans.
This article outlines Kubernetes cluster components and presents practical single‑cluster and multi‑cluster deployment architectures, covering component interactions, load‑balancer designs, pre‑deployment preparations, Ansible‑driven automation, and operational benefits for private‑cloud PaaS environments.
Linkerd 2.8 adds a multi‑cluster extension that securely connects Kubernetes clusters across any network topology, introduces an add‑on plugin system with Jaeger and Grafana support, and includes numerous performance and stability improvements for cloud‑native service mesh deployments.
This guide explains how to set up Istio across multiple Kubernetes clusters using a multi‑control‑plane topology, covering CA secret creation, control‑plane installation, DNS configuration, sample application deployment, ServiceEntry and DestinationRule creation, and cross‑cluster traffic routing verification.
This article explains the challenges of operating many Kubernetes clusters in private‑cloud environments and presents a declarative, operator‑driven Kube‑On‑Kube architecture that dramatically cuts deployment, upgrade, and user provisioning time while remaining cloud‑native and infrastructure‑agnostic.
This article explains why Kubernetes was chosen, describes the overall architecture, high‑availability master design, multi‑IDC cluster deployment, logging, monitoring, service exposure, image building, lifecycle hooks, CI/CD, multi‑cluster management, encountered challenges, and future plans for operators and automated scaling.
This article analyzes the complexities of deploying Kubernetes on Alibaba Cloud, compares bare‑metal and virtual machine worker nodes, evaluates container runtimes and security sandboxes, and provides practical guidance for selecting instance types, cluster size, multi‑cluster strategies, and serverless options.
This article explains why Youzan chose Kubernetes, describes their multi‑IDC, multi‑cluster architecture with high‑availability master components, logging and monitoring solutions, custom service exposure, image building process, lifecycle hooks, continuous delivery pipeline, operational challenges faced, and future plans such as operators and auto‑scaling.
This article explains the rise of multi‑cluster designs, outlines three common scenarios—cloud burst, disaster recovery, and active‑active—examines the complexities of application delivery across clusters, and details how Kubernetes and Alibaba Cloud’s ACK implement unified APIs, tunnel mechanisms, and high‑availability to enable true multi‑cloud operations.
The article summarizes the 58 Group technical salon where Zhihu and 58 teams shared their cloud platform evolution, containerization strategies, multi‑cluster management, network architecture, service discovery, and real‑world case studies, highlighting challenges and solutions for large‑scale Kubernetes deployments.
Didi transformed its massive single‑cluster Elasticsearch deployment into a transparent multi‑cluster architecture using TribeNode and cross‑cluster search, isolating workloads, reducing fault impact, and achieving five‑fold scale while preserving a single‑cluster appearance for services, despite added configuration complexity and stability challenges.
This article details Suning's approach to overcoming HDFS Namenode performance bottlenecks by partitioning into multiple clusters, leveraging Alluxio's unified namespace, and presenting design decisions, implementation challenges, and performance test results that show significant throughput and latency improvements.
