At one point during the 3.2 development cycle, we saw 2d geoNear queries micro-benchmarks performed at a higher throughput level (~10% better). This ticket is filed so that we can see if there are ways we can get back to the peak level that we have seen for these operations.

As a target we can use git:7f1b07c and the geo.near.2d.find100 workload in mongo-perf suite for cross comparison and analysis purpose.