Skip to content

Adoption & Ecosystem

Who uses it

The repository's ADOPTERS.md carries no names; it points to the official adopters page and issue #4540. The first-party source for named adopters is the Karmada Adopters page, which lists more than 40 organizations. A sample, all cited from that page (accessed 2026-06-24):

OrganisationUse caseSource
Alibaba CloudListed adopterkarmada.io/adopters
BloombergListed adopterkarmada.io/adopters
DaoCloudListed adopterkarmada.io/adopters
HuaweiListed adopterkarmada.io/adopters
iQIYIListed adopterkarmada.io/adopters
ShopeeListed adopterkarmada.io/adopters
Trip.comListed adopterkarmada.io/adopters
VivoListed adopterkarmada.io/adopters
WellhubListed adopterkarmada.io/adopters
ZTOListed adopterkarmada.io/adopters

The adopters page does not describe per-company workloads, so the use case column reflects only listed-adopter status. The project's origin also names co-initiators First Automobile Works, ICBC, SPD Bank, Qutoutiao, VIPKid, and xiaohongshu (CNCF blog).

Adoption signals

  • GitHub: 5,503 stars, 1,149 forks, 746 open issues, 71 watchers (GitHub API, 2026-06-24).
  • Contributors: the GitHub API returns 307 contributor accounts (2026-06-24, includes bots). At incubation in 2023-12, CNCF reported 500+ contributors from 60+ organizations across 20+ countries with 7 maintainers; the two figures use different counting and are not directly comparable (CNCF blog).
  • Release cadence: v1.18.0 is the recent stable release (2026-05-30), with v1.19.0-alpha.0 in progress (GitHub API).
  • A formal Adopter Group program launched in 2025-03 (CNCF blog).

Ecosystem

  • GitOps: ships interpreter customizations for Flux and Argo, so it can run alongside a GitOps delivery tool.
  • Cross-cluster networking and discovery: multi-cluster Service (MCS) for service discovery across clusters.
  • Autoscaling: FederatedHPA and CronFederatedHPA scale across clusters, backed by karmada-metrics-adapter.
  • Cross-cluster query: karmada-search provides search and caching over all member clusters.
  • CRD workloads: third-party interpreters cover Flink, Ray, and Kubeflow CRDs so they can be propagated and replica-split like native workloads.

Alternatives

Karmada keeps Kubernetes-native templates and adds independent propagation/override policies plus dynamic cross-cluster scheduling and replica splitting, which puts it on the automation-heavy end. Open Cluster Management leans toward placement plus policy and governance. Fleet drives delivery from Git and is weaker at progressive rollout (CNCF blog, 2022-09-26).

AlternativeDiffers by
Open Cluster ManagementGovernance and policy oriented; foundation under Red Hat ACM.
Rancher FleetGitOps-driven delivery with label targeting; weaker progressive rollout.
ClusternetCNCF Sandbox multi-cluster management, smaller scope.
KubeFedDeprecated federation v2; Karmada is its successor.