From: Maya Chen <maya.chen@example.com>
To: Storage Engineering <storage-eng@example.com>
Subject: Follow-up: elevated write latency in the key-value store
Date: Thu, 30 Jul 2026 14:18:00 -0400

Hi team,

We traced this morning's write-latency spike to two overloaded partitions in the us-east cluster. A recent tenant migration shifted enough traffic to make their leaders CPU-bound, while follower replication remained healthy.

We rebalanced the affected partitions and p99 write latency returned from 420 ms to 38 ms. No acknowledged writes were lost, but approximately 0.6% of requests timed out and were retried by clients.

For the permanent fix, I propose lowering the automatic rebalance threshold and adding an alert when leader CPU exceeds 75% for ten minutes. I'll open the changes for review this afternoon.

Thanks,
Maya