WiredTiger currently tracks statistics that let us compute the total cache miss rate as:

miss rate = wt cache pages read into cache / wt cache pages requested from the cache

In order to better understand some of the trade-offs of changing leaf page sizes, it would be useful to see the cache miss rates for those different page types.

This ticket should add statistics similar to the above that separately count internal and leaf pages read into the cache, and the number requests for each page type from the cache.

The initial use case is to support evaluation in some ad hoc performance testing. If we decide to keep these metrics in the long term, we should separately add derived metrics for the internal page miss rate and the leaf page miss rate to t2.

CC jagadish.nallapaneni@mongodb.com