The current workload doesn't stress 4.2 as much as we'd like.

The first thing we've noticed is low cache usage across the board which is likely due to the fact that the dataset itself isn't large relative to the available cache size.

We should experiment with increasing the number of documents/collections, document size, etc.