If you run multiple rm -r commands in parallel, things get done MUCH faster. Your filer can handle the load no problem and chances are your unix host can easily handle at least 10 rm processes and probably quite a few more, depending on number of processors, etc.
Even on a single CPU unix host, you see dramatic speed up because when one rm process waits for a NFS reply from the server, it relinquishes the CPU to another rm process that is ready to resume running because it's NFS reply has arrived.
Of course, it helps if your data is split up somewhat evenly in several directory trees.
for dir in /some/where/*; do /bin/rm -r $dir & done
I just wish I could delete files on a NetApp from the console. It takes forever and a day to remove a hundred or more GB off a filer from either CIFS or NFS.
Steve Losen scl@virginia.edu phone: 804-924-0640
University of Virginia ITC Unix Support