Note: this is in prototype development stage and is not yet functional

Disk access (even if mmapped) is slow primarily based on the number of disk seeks required to access data.

While mongo mmaps data into RAM that only provides a speedup if your data is already in RAM, if it's on disk it doesn't help. When you try to scan say 10k records in a mongo query, it must perform disk seeks on both the index and the data extents to complete a query. This creates a significant cold start problem.

mongosort attempts to sort ondisk data ordered by the primary key _id so that when using custom _id values and querying based on that sort order, it takes as few seeks as possible to map data in from disk.

http://2013.nosql-matters.org/bcn/wp-content/uploads/2013/12/storage-talk-mongodb.pdf https://speakerdeck.com/mathias/storage-internals