SERVER-111573 will allow more wiredtiger data source statistics to be queried in mongodb. These data source metrics about the b-tree could give insight into the file layout of a collection/index. Here are a few more stat suggestions on how we could extend these:

• maximum leaf page seen - this would allow use to see if there's an abnormally large block in the file. For context, pages can grow beyond the configured 'leaf page max size' if the value size is bigger than the page size. This is to avoid overflow pages. Large pages may have consequences in efficiently reusing space. This would be different to the existing 'maximum leaf page size' as that tracks the configured leaf page max size, not the max size seen in practice.
• block/page size distribution
  • buckets - similar to the compression ratio histogram stats.
  • (Yury) minimum and maximum offsets, normalized block positions of them and perhaps average, minimum, and maximum leaf page depths