Ctrip Technology

Jul 10, 2026 · Cloud Native

How Ctrip Scaled Karmada to Handle 200 GB Memory Peaks and Hundreds of Thousands of Pods

This article details Ctrip's migration from Kubefed to Karmada for multi‑cluster governance, describing the architectural evolution, production deployment for high‑availability and smooth cross‑cluster pod migration, and the performance and memory optimizations required to support over 200 GB of control‑plane memory usage and tens of thousands of Pods.