Architect
Feb 18, 2022 · Cloud Native
Large‑Scale etcd Cluster Performance Optimization and Pod Data Splitting in Ant Group’s Sigma
This article describes how Ant Group tackled the performance ceiling of its massive Sigma Kubernetes clusters by horizontally splitting etcd storage for Pods, Leases and Events, redesigning watch handling to avoid component restarts, and using snapshot‑based migration to preserve data integrity while reducing latency.
Data MigrationKubernetescloud-native
0 likes · 27 min read