Generally speaking, is there a way to find out just how busy all Mongo instances really are (within the cluster) so that I could potentially scale down the number of instances that we have? I understand that Inserts/updates/deletes happen across the cluster (on the primary…then reflected into the replicas) and that our reads are pointed towards the replicas (as balanced by mongos). Are there some internal commands (in the data-dictionary) that I can use for this or do I have to try to go by OS/network metrics? One underlying question that I do have is how mongos load balances the client connections (that are reading data) to particular nodes; When I look at “network Out” in cloudwatch on all 7 ec2 instances, I can see 2 or 3 nodes that are really low in comparison to the others…which seems like maybe only 5 are predominantly used?