I have used Patch/iX about half a dozen times on our
957 and 928.  Its main, and very significant advantage
is the ability to do most of the prep work while the system
is up and available to users.  Actual down-time on both
machines generally involves only a little over an hour
to boot from the consolidated CSLTs generated by Patch/iX
and then do Phase II (restoring subsys products off the
CSLT and a few install jobs, including the always necessary
JCONFJOB).

A few minor problems:

The patch qualification process needs to be followed very
carefully.  I know its tedious, but you just have to do it.
I always take a patch that automatically qualifies.  If one
doesn't qualify, I always read the notes on why (usually
because it has already been installed by a previous process,
or I don't have all the parts of the product it proposes
to patch).  I generally FORCE qualification of new functionality.
I've done this often enough to get really annoyed at the
inconsistency of Patch/iXs cursor and page movement controls.
Its infected with UNIX, unfortunately.

There is a point (I forget exactly where, but I've complained
about it in the past and got no meaningful response) where
a redundant restore of subsystem product files will occur
(obviously only if the process involves a subsystem tape).

The Phase II process doesn't clean out its history file,
so next time (probably months later) when you fire it up
again, you need to say 'NO, I don't want to restart the
previous process'.  THEN, it clears out its logs.  That
sure makes you think ... did I finish it up last time?

I'm confident with this method of updating--despite the
little problems.  It is VERY MUCH improved over the old days
when you had to spend a high-pressure 4 to 6 hours down
time.

FWIW,
Ron Burnett
Women's & Children's Health Care Network
Melbourne
[log in to unmask]