How Upgrading EBS Volumes Boosted etcd Write Performance by 30%
A technical deep‑dive shows how a team managing dozens of Kubernetes clusters diagnosed a write‑ahead‑log bottleneck in etcd, measured IOPS and latency with etcdctl and fio, upgraded gp2 volumes to gp3, and discovered diminishing returns beyond 3000 IOPS while explaining the role of fdatasync in storage performance.
