During our daily DRM process, we backup the storage pool to tape.  In
the last two weeks, we've nearly doubled the number of tapes we
generate.  How can we find out who are the big users?  I see no total
usage stats maintained at the filespace level (too bad).  How do I write
QUERY/SELECT on all files (which number in the hundreds of thousands),
grouped by node, without it taking forever to compute?  Am I better off
dumping the query output to disk and then using something like SAS to
process?  Any ideas anyone?