In WT-12398 I added some backup-related statistics. One statistic, backup_blocks, counts the number of blocks changed for an incremental backup. It would be much more useful to have other or different statistics to be able to see how many blocks changed with a collection file versus an index file for MongoDB.

However, WT does not know anything about those files. One potential path that keeps that separation would be to have statistics that separate out on a table property inside WiredTiger that is different between collections and indexes that WT can know.

Some potential properties are:

• Whether or not compression is configured on a table. Collections use compression, indexes do not.
• memory_page_max is 10M on collections and default (5M) on indexes.
• leaf_page_max is 16k on indexes and default (32k) on collections.
• internal_page_max is 16k on indexes and default (4k) on collections.

Of these, it seems easiest to justify the compression one. The changed blocks represent chunks of file on disk. Compression impacts data on disk.