You can easily determine how many "lines" of tapes you will initially
break out into, based upon which of the multiple types of collocation
you ultimately decide upon and how many nodes or filespaces are
active, sending data to TSM storage.  I would double that number to
account for "tape leak", based upon historic experience.  And if there
are Windows clients, think more about DIRMc in that context.  I think
most of us implement collocation judiciously, based partly upon
library cell impact, but probably more so on realities of tape drive
contention.  A big arrival disk pool facilitates more realistic
collocation.

   Richard Sims