Just a few words, as I 've other some things to do. I've just published a development version of mongodb-statistics on github.

This module is a first try on replacing heavy MySQL operations done by the core statistics module

It's copy-and-alter of the core statistics module, with some additions. So it's not the nicest piece of code I could write :-)

On the things added :

• batch migration of previous node_counter statistics if any (not yet for accesslog)
• post-synchronisation of node_counter mongodb table to a sql table (via cron), so that you could query it via views (not for accesslog, are you mad?)
• time based caching of the popular content block

Still lot of thing to do (see the TODO) but it was a nice way to learn mongodb. I especially like the idea of mapReduce functions applied for complex GROUP BY equivalents.