Why do we need this?
We are observing a significant difference in the SCATTER_GATHER timer metric value at the broker layer and the TOTAL_QUERY_TIME timer metric value at the server layer.

Additional metrics across each phase at the broker layer would be helpful for debugging it.