The discussion about limiting testing in production LPARs, and the slight 
sidetrack I went down with the DB2 governor got me thinking about all the other 
limits we can impose on production work.

Some examples that can be enforced by system exists or monitoring products:

-- CPU time limits enforced for batch jobs
-- Wait time limits
-- Spool output limits
-- Memory limits for batch jobs, started tasks, and/or TSO users
-- Transaction CPU limits (DB2 governor & CICS I think now has something 
similar, although some monitors have done so for years)
-- CPU parallelism (or not) in DB2
-- WLM resource limits
-- LPAR hard caps / soft caps
-- Others?

In my previous position I took care of most of the system level ones and over 
time they were increased. But we always had most of the above in place. I'm 
curious if anybody has taken the approach of eliminating some of those limits 
in production. Or maybe never had them. Do you allow production batch jobs to 
specify unlimited CPU time? Have you taken out IEFUSI and let everything use as 
much virtual storage as they want? 

As system configurations get larger, it makes sense that the limits would at 
least increase over time. But should they be eliminated? I'm not so sure about 
that, so I'm curious about the approaches others have taken.

Scott

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to