There's the usual answer around IO performance.  Avoid Raid 5, separarte
your data from logs, and separate your nonclustered indexes.  One thing you
might test is on any table that needs the clustered index defraged, drop the
nonclustered indexes, defrag, then rebuild the nonclustered indexes

Danny hit the nail on the head.  You need to place the log files on a
separate Raid 1 by them selves for peak write performance.  And a RAID 5 is
a dog when it comes to writes.  You should also try DBREINDEX instead of
DEFRAG to test the differences.  Just make sure you have plenty of free
space in both the data and log files before executing the command.

The most efficient way to improve the performance of a step is to eliminate
the step entirely or at least minimize the frequency that it occurs. When
bulk loading a table, try ordering the select to match the clustered index,
and perhaps this will reduce index fragmentation. You may also want to
consider dropping all indexes before the bulk load and then re-creating them
afterward. This will make fresh indexes in perhaps the same amount of time
and should also improve the performace of the bulk load since indexes are
not maintained as new rows are inserted.