Interactive Feature -vs- Total System CPW (%CPU)

 I have a customer this is maxing out his Interactive Feature.  I want to
ensure that if he upgrades this feature he will come out of his performance
problems.  The problem I have is separating the interactive %CPU from the
total system %CPU.  This is because when he reaches the interactive capacity
threshold the dammed CFINT0n tasks kick in and the total system craps out.
I know all about what IBM kindly calls "speed bumps" are and how they work.
I prefer to call them "curbs" as in how the automatic breaks are applied to
the stock market.

 I have looked at performance reports but don't see a clear definition of
total system -vs- interactive when CFINT0n is consuming 90+% CPU.   The
bottom line is I don't want them to spend big bucks on the interactive
upgrade and then find out that the total system is under powered.  My review
saw no problems with processors or memory, and DASD usage is fine.  I'm
suggesting a number of tuning adjustments but they are likely to only smooth
out some of the current "feel" and not expected to fix the big picture.
Summer is their peek season and they'll die if this is not resolved very
quickly.

 I know about the 2 SYSVALS QDYNPTYADJ & QDYNPTYSCD.  They have them set on
(1) and I'm going to have them turned off (0).  I'm also considering using
FAST400 as a diagnostic tool to help define the problem.  However FAST400 is
not a solution (many issues, don't ask) and the client is willing to pay the
IBM price for the upgrade if that's the fix.  They will investigate a full
replacement to a new 825 but business issues prevent it at this time.

 The system details are:
 720-2064-1504 4-way 6Gig mem CPW 1600/560.
 The next interact step to a 1505 changes 560CPW up to 1050CPW or roughly
from 1/3 to 2/3 of the total system.

 With the cost of the upgrade being almost $100k this decision must be
backed by proven evidence.   If the total system is truly being maxed out
this is not the correct solution.

 Your comments please.