[mej10]

We primarily use it to aggregate a large-ish (10 TB/day) amount of data for insertion into an analytics database.

The code is very straightforward and it is fast.