It would be great if Mongo could update a Map Reduce Collection every x minutes. This was M/P functions that would be slightly inefficient to run on the fly, could be run in the background and the application itself would just need to read the map reduce result collection. This would work tremendously well for Map Reduce results that dont change in real time, but more so in pseudo-real time.

e.g. authors -> name, books_per_year, avg_pages_per_book, foo_metric, bar_metric, ..
book -> name, author , ..., subject, date_published
Assuming 1 million authors and 10 million books

M/R to get "all authors whose foo_metric is between x & y and books_per_year is between x' & y' and date_publihsed is after 1-1-1990" would be very inefficient currently, but could be "refreshed" every x mins. by mongo.