We mainly capture GC metrics and alert on them. One good thing that happened is there is no longer GC related alerts happening in production anymore. Also tail latency for API calls from kibana to ES improved.