Wikistats are now available for all Wikimedia projects.
Stats have been generated from the latest dumps, which are from March 9.
Wiktionaries (use refresh button when you see old stats)
If there were a machine dedicated to stats of various
with its own mirror of the db, could this be done more efficiently
as a running-total, updated whenever the db was updated?
A dedicated mirrored server for these and other background jobs (e.g. IBM
history flow charts and Special Pages) would certainly help. Switching away
from processing dumps towards direct SQL access has been suggested before.
Especially with the upcoming database scheme change, dumps become more
difficult to handle, though not impossible.
I had my doubts about switching, because very heavy 'Special:' SQL queries
have been often disabled or cancelled halfway to bring down server load, and
the stats queries will be very heavy indeed.
A separate server would make this less of an issue. Average CPU speed will
do fine, memory no less than 1Gb, preferably 2GB. Even though I took great
care to keep memory requirements at an acceptable level, some hash tables
need to be in memory and their sizes grow like everything else around here.
I hope you don't mean running-total literally, where all figures change
every second :)
At least it should be possible to generate new stats once or twice a week,
instead of once a month (the latter depending on how often dumps are