This incident was resolved.

The root cause was a runaway DB instance that caused higher commit latencies. Our engineers have a plan on mitigating this going forward