Goal / Product Question:

Our recommended topology is a replica set with three nodes for high availability. However, some customers are questioning the cost associated with running all three nodes, where they claim reads & writes only happen on their primary.

We offer the ability to redirect reads to a secondary node. We believe there is a population of customers who use secondary reads, but we never had a good way to identify them. Exposing a metric to identify if a secondary node has served reads will help identify these customers who specifically directed reads to a secondary node.

Must have: 

-Number of clusters with secondary reads

-Distribution of readPreference options (e.g. secondary, secondaryPreferred, primary)  

Nice to Have:

-Hedge Reads 

-Mirror Reads

-Number of clusters with reads with multiple read preferences