What is Kubernetes multi-cluster management?

Multi-cluster Kubernetes becomes necessary for: geographic distribution (low latency for global users), fault isolation (one cluster failure doesn't affect others), regulatory compliance (data sovereignty), environment isolation (prod, staging separate clusters), workload isolation (different security domains). Challenges: service discovery across clusters, cross-cluster networking, consistent configuration, security (different API servers, RBAC), unified observability, GitOps at scale. Tools: (1) Argo CD: single Argo CD instance managing multiple clusters. ApplicationSets with cluster generators: generators: - clusters: selector: matchLabels: environment: production. Add clusters: argocd cluster add my-cluster; (2) Flux: bootstrap Flux in each cluster pointing to the same Git repo with cluster-specific overlays; (3) Cluster API (CAPI): Kubernetes-native cluster lifecycle management. Manage cluster creation/upgrade/deletion via Kubernetes CRDs on a management cluster. Providers: AWS (CAPA), Azure (CAPZ), GCP, vSphere; (4) KubeFed / Admiral: federation-level resource management (deprecated in favor of newer approaches); (5) Submariner: cross-cluster networking — connects pod/service networks across clusters securely. Service mesh multi-cluster: Istio multi-primary or primary-remote topology — services discoverable across clusters via service mesh; (6) Karmada: multi-cluster orchestration — propagate policies to multiple clusters, cluster failover. Platform teams: internal platform built on Kubernetes (Backstage + Crossplane/Argo CD) gives developers a self-service portal while abstracting multi-cluster complexity.