I'm trying to find an objective test to find if a backup bottleneck is with
the disk system, the tape hardware or somewhere else in the chain. TSATEST
is supposed to do that. However, I've been unable to find a baseline
performance that can be considered good or bad.

Ofcourse it all depends on the harddisks, RAID controllers and the system
itself, but it would be handy to have a guideline saying something like 'Use
TSATEST with blocksize <x>, with <y> threads and such-n-such other options
to get a realistics performance test for Arcserve'. Similar guidelines for
Syncsort, Legato, NetVault, you name it. Does Novell offer such a list??