Abstracting Multi-Cluster Topologies With CAPI and Linkerd for Internal Developer Platforms

on YouTube ↗18m

" summary

Platform teams use Cluster-API (CAPI) to declaratively provision multiple Kubernetes clusters and run Linkerd in each cluster to create Linkerd federated services that present a single endpoint to developers. A multicluster topology and Coordinant/Backstage APIs hide cluster topology while federated services detect failed clusters and route traffic only to healthy cluster instances, preventing 500s. The demo shows federated services built across three clusters via CAPI and Linkerd and the failover behavior working as described.

AI-written summary · about

" chapters (9)

1. 0:00Introduction
2. 1:05Resilience in platform engineering
3. 2:53Resiliency as platform capability
4. 6:05Linkerd for multi-cluster resiliency
5. 7:32Traditional multi-cluster challenges
6. 9:16Federated services with Linkerd
7. 11:51Hiding cluster topology
8. 13:59Abstraction vs developer awareness
9. 14:20Closing remarks

" related talks

• 28mMulti-Cluster is Easier Than You Think with Linkerd and AmbassadorThomas Rampelberg · KubeCon + CloudNativeCon North America 2020 Virtual
• 12mService Mesh & Kubernetes Multicluster Communication With LinkerdMohammad Akif · KCD Dhaka 2024
• 09mMulti-Cluster Networking: A Technology-Agnostic Network Operator for K8 FederationRyan Jenkins · DevConf.CZ 2025
• 25mIntroduction To Cluster APIJussi Nummelin · Cloud Native University North America 2025
• 36mKubernetes Multi-Cluster Made Easy - Federica Ciuffo and Marco BalleriniFederica Ciuffo, Marco Ballerini · ContainerDays 2023
• 32mCluster API as CodeDavid McKay · KubeCon + CloudNativeCon Europe 2021
• 06mKeynote: From One to Many, the Road to MulticlusterKaslin Fields · KubeCon + CloudNativeCon North America 2021
• 01h 31mMulti-cluster Failover Using LinkerdCharles Pretzer · KubeCon + CloudNativeCon Europe 2022